You can subscribe to this list here.
2004 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
(68) |
---|---|---|---|---|---|---|---|---|---|---|---|---|
2005 |
Jan
(21) |
Feb
|
Mar
(1) |
Apr
|
May
|
Jun
(1) |
Jul
|
Aug
|
Sep
|
Oct
|
Nov
(1) |
Dec
(3) |
From: MaDProph <mad...@la...> - 2005-12-06 10:38:50
|
=20 Hi Lukas =20 Yeah, still developing here, but we've slowed down due to our personal lives. Actually, we're even planning a final 0.8 release (some bug = fixes, updated docs & some translations) for mid December, so nice timing here = ! =20 First of all, the team always welcomes motivated people. If you've got = your own ideas the better ! =20 About your idea : I understand it'd involve the production of some sort = of report of your file database. Your idea is a very good one, in the = spirit of MediaLibrary (being a "file collector"). NB. Functionally it is between = a report (human readable statistics or listings), an export (to another = data format) and a save (to another supported catalog format - which means it would improve ML's openness); this flexibility is a good thing, but we should think about ways of simplifying it for the user by separating = this three functions (even if there's almost the same tech below, read on). =20 Actually, there's already the "HTML report" function, which allows to = write an HTML file with all files in your catalog OR all categories under a = given category (which is configured in resources/medialib.prop) with some = computed aggregate information (like category size & file number). However it is = but a very crude beginning of an implementation of an actual report. =20 I think there are two ways to do a better job in ML : =20 * if you need tight integration with the GUI (for example for tuning the report format) and weird information computing, go the java way and use = some open source report generation java library (plenty of them). =20 * if you need flexibility in the report you're generating, go the XSL = way and write an XSL stylesheet which will translate the xcat format to the = open format of your choice (styled HTML, or open office as you say etc.). = This flexibility would be a must. However this function would require ML to = save the catalog before each reporting, therefore it would require more time; moreover the report formatting would not be editable in MediaLibrary = since it would be in XSL (save if MediaLibrary would be able to tune some = specific XSL generation, say for styled HTML). =20 In either case MediaLibrary would be able to view anything produced in = html thanks to the builtin browser. =20 If I'm allowed to speak my mind, I think the second is still the easiest = and more flexible way to go. The disadvantages are minor in the context of a report generating functionality. It goes the way you're speaking as = well. =20 Technically : The second option would imply to integrate * an XSL engine into MediaLibrary (easy, done that before) * a dialog box to locate both the XSL and the output (trivial) * an optional call to the browser at the end of report generation (easy = to do in the ML framework) =20 There are many ways to improve the functionalities from there. For now, = what do you think about it ? Ever done XSL ( <http://www.w3schools.com/xsl/> http://www.w3schools.com/xsl/) ? It's actually not that hard (especially with the xcat format) and quite useful ! If necessary, I'd be willing to = do some coaching here and help you start the thing up... =20 =20 Don't worry, if you prefer doing a "companion tool" in c#, we'll welcome = it as well since it would nevertheless be a tremendous increase to the = whole package. However it'd mess with the multiplatform idea, and be less integrated and less potent. Anyway we'd open the CVS & all to you. =20 I'd like to hear your thoughts about all that. =20 MaDProph, for the ML Team =20 =20 -----Message d'origine----- De : koc...@po... [mailto:koc...@po...] Envoy=E9 : mardi 6 = d=E9cembre 2005 01:40 =C0 : mad...@us...; = od...@us... Objet : medialibrary improvement idea =20 hi, medialibrary is really good software! what happened to you? are you = still developing? =20 i'm thinking about implementing new tool (it could be even very useful = for me and it's very easy to make it): =20 (minimal requirement:) ------------------------ -you'll choose some category and all the directories (or files) = appertaining to it will be: a) saved into text file b) saved into pdf file (or at least .tex file) =20 -working only for .xcat files (xml is very clean and easy to use) -implemented for .NET.. ------------------------ =20 or, add this function into your project. I don't know java as good as c# = , but It shoudn't be big problem to learn :) it can be further improved, function of saving into .ods (file for Openoffice.org "Calc" (equivalent = for EXCEL in MS office)) and you've got your cd collection in the table! =20 I can join your team (and add this functions - or you can make it), or = make it as my own application for .net. Tell me what do you think about that. next week I'm leaving abroad, but in the end of January, I could (teoretically) start to work. =20 Lukas, Czech Republic =20 =20 =20 |
From: MaDProph <mad...@la...> - 2005-12-06 10:34:59
|
Yesssss that's it !!! Great ! My most sincere thanks to you - Italia, here we come ^^ It looks all right, I'll test it tonight. Of course, you'll be listed = among the contributors ! Besides we'll need to test the new version before = putting it on sourceforge - mainly 10 minutes basic testing, like making sure it launches on one's platform & no functionality among the main ones = crashes... would you like to help ? This way you'd have the release before = everybody else ;) Anyway many thanks once more for your work. Oh, and I forgot to reply to your praises of the new ML version. So = thank you as well, this actually is valuable feedback in our quest toward = improved user experience - and some welcome moral boost ^^^ Later, MaDProph -----Message d'origine----- De=A0: Roberto Mannai [mailto:rob...@po...]=20 Envoy=E9=A0: mardi 6 d=E9cembre 2005 09:39 =C0=A0: MaDProph Objet=A0: Re: medialib Hi, I don't know what may be happened. Anyway, I re-attach the file, let me=20 know wether it arrive correctly or not, please. Ciao, Roberto MaDProph ha scritto: >Hi Roberto > >Thank you very much for your contribution. However, the only attachment = I see=20 >in your mail is a "MessageBundle_it_IT.properties" file but with the content=20 >of the "about_IT.html" file ! Either you've mixed up the names, or you *gasp*=20 >forgot to translate the actual MessageBundle.properties file ;_; > >If you're still willing to do it (I sure hope so), I advise you to go = in the=20 >"resources" directory copy, and there copy the MessageBundle.properties file=20 >to a MessageBundle_it_IT.properties, and translate this one to italian = (see >the attached TRANSLATION_HOWTO updated, along with the original=20 >MessageBundle.properties). When starting MediaLibrary 0.8pre, you = should=20 >actually see everything you're translating on the menus, buttons, = tooltips=20 >etc. > >The file you've already translated should go in the "about_IT.html" = file in >the "resources" directory and be displayed when clicking on the help > about=20 >button in the menubar. > >Besides, thanks for the bug report, it rings a bell somehow; I'll make = sure it=20 >doesn't creep in the final release (before mid december) ! > >later, MaDProph > >Le dimanche 4 D=E9cembre 2005 23:44, vous avez =E9crit : > =20 > >>Hi, >>I've finished to translate the italian version :) - see attachments. >>I like very much this version!!!! I like above all the improvements on >>searchs and filters, with and / or pipelines, as well as = cross-category >>search. >> >>Just a note about third tab, "Search", with "show the search options": >>if I type "pippo.mp3" (which is of "MUSIC" category) and I add only = the >>category "VIDEO" in the box, why the result contains pippo.mp3? I was >>guessing that the text and category go in AND. >>The search on Statistics, on the contrary, does not return pippo.mp3, >>like I would expect (using the same criteria). >> >>Anyway, great tool, congratulations :) >>Ciao >>Roberto >><cid:par...@po...> >> =20 >> >>-----------------------------------------------------------------------= - >> >># Default Message Bundle for MediaLibrary >> >># Filter >>filter.bak_description =3D Backups >>filter.cat_description =3D ADC Catalogs >>filter.???_description =3D=20 >>filter.xcat_description =3D XML Medialib Catalog >>filter.xzcat_description =3D Zipped XML Medialib Catalog >>filter.html_description =3D HTML files >> >># Menus >>menu.file =3D File >>menu.index =3D Index >>menu.edit =3D Edit >>menu.tools =3D Tools >>menu.help =3D Help >> >># Actions >>action.about =3D About >>action.about.htmlfile =3D resources/html/about.html >> >>action.help =3D Help >>action.help.htmlfile =3D = doc/o3handbook/html/MediaLibrary_User_Guide.html >> >>action.exit =3D Exit >> >>action.trace =3D Display the logs >>action.trace.title =3D Trace Logs >>action.trace.refresh =3D Refresh >>action.trace.filenotfound =3D The file medialib.log does not exist. \n Cannot load the logs. >> >>action.index =3D Index >>action.index.defaultmediumname =3D New Medium >>action.index.cantfindmedia =3D Can't find medium >>action.index.notmedia =3D is not a medium path. >>action.index.indexingerrors =3D Indexing errors >>action.index.emptyfolder =3D is an empty folder ! It may be an empty = floppy. >>action.index.recursivescanning =3D Be carefull, probably recursif = directory scanning ...\nIf it is not the case, increase the alert threshold in the preferences and try again. >> >>action.load =3D Load >>action.load.nocodecfound =3D No codec found for file >> >>action.close=3D Close >> >>action.move =3D Move >>action.move.enterparent =3D Please enter the path of the new parent. >> >>action.new =3D New >>action.new.entercategoryname =3D Please enter the new category name. >> >>action.openbrowser.urlerror =3D Unable to load URL >> >>action.preferences =3D Preferences >>action.preferences.cantload =3D Can't load preferences file! >>action.preferences.errorempty =3D Error: empty preferences! >> >>action.remove =3D Remove >>action.report =3D HTML Report >> >>action.save =3D Save >>action.saveas =3D Save As... >>action.saveas.specifyextension =3D Please specify a file extension = first >>action.saveas.nocodecfound =3D No codec found for the file extension >>action.saveas.askoverride =3D You're trying to save over a different existing file!\nOverride anyway? >>action.saveas.overrideattempt =3D Override Attempt >>action.saveas.bakforbidden =3D Can't save to a backup file. >>action.saveas.creationerror =3D Unable to create file >>action.saveas.notfilepath =3D is not a file path. >>action.saveas.notwriteable =3D is not a writeable file.\nAborting = save. >>action.saveas.abortingsave =3D Aborting save. >>action.saveas.technicalerror =3D A technical problem has occured. = Please report it to the Medialibrary team. >> >>action.search =3D Search >>action.search.title =3D Find a File >>action.search.tooltip =3D Enter the file searched. It behaves like = google \ >> (spaces between words are considered as AND operators) >>action.search.options =3D Options >>action.search.options.show.tooltip =3D Show the search options >>action.search.options.hide.tooltip =3D Hide the search options >>action.search.options.casesensitive =3D Case sensitive >>action.search.options.comments =3D Look in comments >>action.search.options.categories =3D Categories >>action.search.options.categories.tooltip =3D Add the categories you = want to be required >>action.search.help =3D Help >>action.search.helptitle =3D Help on Search >>action.search.advanced =3D Advanced Search >>action.search.results.title =3D Results >>action.search.results.tooltip=3DSort the result rows by clicking on = the column headers.\n\ >> Click on a file to select it in the media tab. >>action.search.results.file =3D File >>action.search.results.media =3D Medium >>action.search.results.path =3D Path >>action.search.results.relevance =3D % of Relevance >>action.search.results.nofilefound =3D No File found >>action.search.results.toomuchfiles_1 =3D More than >>action.search.results.toomuchfiles_2 =3D files have been found. \n\n\ >> Please refine the query of your search. \n\ >> For example use multiple words separated by blank spaces.\n\n\ >> Note : no displaying such amount of files is a feature, not a = limit. >>action.search.files.title =3D Files >>action.search.files.tooltip =3D Displays the files found. >>action.search.statistics.title =3D Statistics >>action.search.statistics.tooltip =3D Displays statistics on the file = found. >>action.search.filtered =3D Filtered search >>action.search.filtered.tooltip =3D Trigger filtered search mode >>action.search.filtered.backtobasic.tooltip =3D Go back to simple = search mode >>action.search.filters.title =3D Advanced filtering >>action.search.filters.pipelines.title =3D Liste of parallel pipelines >>action.search.filters.filterType =3D Type of filter to be created : >>action.search.filters.googlename.description =3D File Name >>action.search.filters.googlename.casesensitive =3D Case Sensitive >>action.search.filters.googlecomment.description =3D Comments >>action.search.filters.googlecomment.casesensitive =3D Case Sensitive >>action.search.filters.category.description =3D Categories >>action.search.filters.category.acceptsubcategories =3D Also = Subcategories >>action.search.filters.category.acceptsubcategories.tooltip =3D Check = to include the subcategories in the search >>action.search.filters.filenode.description =3D Directories >>action.search.filters.filenode.acceptsubfilenodes =3D Also = Subdirectories >>action.search.filters.filenode.acceptsubfilenodes.tooltip =3D Check to include the subdirectories in the search >>action.search.filters.unknownfilter =3D \ is an unknown filter >> >>action.search.filters.pipelines.add.tooltip =3D Add a new filter = pipeline >>action.search.filters.pipelines.remove.tooltip =3D Remove this filter pipeline >>action.search.filters.add.tooltip =3D Add a new filter (to this = pipeline) >>action.search.filters.remove.tooltip =3D Remove this filter >>action.search.filters.filenode.add.tooltip =3D Add a file or directory = to be included in the search >>action.search.filters.filenode.remove.tooltip =3D Remove a file or = directory not to be included in the search >>action.search.filters.category.add.tooltip =3D Add a category to be = included in the search >>action.search.filters.category.remove.tooltip =3D Remove a category = not to be included in the search >>action.search.filters.googlename.text.tooltip =3D Enter the text to be looked for in the file names >>action.search.filters.googlename.casesensitive.tooltip =3D Check to = ensure a case sensitive search >>action.search.filters.googlecomment.text.tooltip =3D Enter the text to = be looked for in the file comments >>action.search.filters.googlecomment.casesensitive.tooltip =3D Check to ensure a case sensitive search >> >>action.selectsubtree =3D Select a subtree >>action.sortby =3D Sort=20 >>action.sortbysize =3D by size >>action.sortbytime =3D by time >>action.sortbyrandom =3D at random >>action.sortbyname =3D by name >> >>action.stats =3D Display Statistics >>action.stats.catalogstatistics =3D Catalog Statistics >>action.stats.chart.categories =3D Categories >>action.stats.chart.filenumber =3D File number >>action.stats.chart.number =3D Number >>action.stats.chart.filesize =3D File size >>action.stats.chart.size =3D Size >>action.stats.chart.yearsfilesize =3D File size over years >> >>action.update =3D Update >>action.update.cantguessmedium =3D Unable to guess updated medium. = Please select it first. >>action.update.askremoveobsolete =3D is obsolete and its data should be removed.\nShould all obsolete files and their data be removed ? >>action.update.obsoletefiles =3D Obsolete files >>action.update.updateaborted =3D Update aborted >> >>action.changecategory =3D Change category >>action.changecategory.categorychoice =3D Category Choice >>action.changecategory.cancel =3D Cancel >>action.changecategory.choose =3D Choose >> >>action.choosefilenode.title =3D File or Directory Choice >>action.choosefilenode.cancel =3D Cancel >>action.choosefilenode.choose =3D Choose >> >># tooltips >>tip.search=3DEnter the name of the file here, as in Google_(c) >>tip.load=3DLoads a catalog >>tip.save=3DSaves current catalog >>tip.close=3DCloses current catalog >>tip.index=3DIndex medium >>tip.stats=3DDisplay catalog statistics >>tip.help=3DBrowse help >> >># inspector >>inspector.name =3D Name >>inspector.category =3D Category >>inspector.comments =3D Comments >>inspector.size =3D Size >>inspector.date =3D Date >>inspector.totalsize =3D Total Size >>inspector.localsize =3D Local Size >>inspector.files =3D Files >>inspector.directories =3D Directories >>inspector.lastscandate =3D Last Scan >>inspector.root =3D Root >>inspector.type =3D Type >>inspector.serialnumber =3D Serial Number >>inspector.mediasize =3D Medium Size >>inspector.freesize =3D Free Size >>inspector.absolutename =3D Absolute Name >>inspector.categories =3D Categories >>inspector.medias =3D Media >>inspector.lastmodified =3D Last Modified >>inspector.commit =3D Commit >>inspector.asksave =3D Do you want to save the changes? >>inspector.nodechanged =3D Node Changed >> >># statusbar >>statusbar.noselection =3D No Selection >>statusbar.selected =3D Selected >>statusbar.objects =3D objects >> >># tabbed desks >>deskpanel.desk.media.title =3D Media >>deskpanel.desk.categories.title =3D Categories >>deskpanel.desk.search.title =3D Search >>deskpanel.desk.statistics.title =3D Statistics >> >># desk statistics >>desk.statistics =3D Compute Statistics >>desk.statistics.tooltip =3D Choose the statistics chart type >>desk.statistics.label.chart=3DChart type >>desk.statistics.label.chart.medialib.desks.statistics.charts.strategies= .Ca tegoryRelativeWeight=3DRelative weight of each category (Pie) >>desk.statistics.label.chart.medialib.desks.statistics.charts.strategies= .Me diaNumberSeries=3DThe media number increase >>desk.statistics.label.chart.medialib.desks.statistics.charts.strategies= .Ca tegoryAbsoluteWeight=3DAbsolute weight of each category (histograms) >>desk.statistics.label.source=3DStatistics source >> >># preferences >>preferences.apply =3D Apply >>preferences.preferences =3D Preferences >>preferences.add =3D Add >> =20 >> >>-----------------------------------------------------------------------= - >> >> >>MediaLibrary Translation HOWTO : >>=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D >> >>NB. the spanish version (es_ES) is taken as an example here. >> >>a. Software messages and labels >> >>Translation does not require any peculiar computer skills :=20 >>All the messages and texts of Medialibrary is externalized in the >>/resources/MessageBundle.properties. >> >>First look for your locale there : http://java.sun.com/j2se/1.5.0/docs/guide/intl/locale.doc.html . >> >>Now the spanish version (es_ES) will be taken as an example. >>To translate, copy this file to a file named MessageBundle_es_ES.properties >>in the same directory, as we did for French with >>MessageBundle_fr_FR.properties. >>Open MessageBundle_es_ES.properties, then translate in portuguese all = the >>Message at the right of the =3D. >>Here "es" =3D spanish language, "ES" =3D Spain country. If you do not = live in=20 >>Spain and wish to make a specific, slightly altered spanish = translation=20 >>available in your own country, copy this file to another file named = for=20 >>example MessageBundle_es_AR.properties (es =3D >>spanish language, AR =3D Argentina country). >> >>If you need to write an end-of-line, please write it like that : "\n". = If you need to write a space between word, write first a backslash, so it = looks like that : "\ ". If you need to write accented characters, please use = the UTF-8 encoding for latin-1 (look for it there : http://www.unicode.org/charts/PDF/U0080.pdf and in the = MessageBundle_fr_FR file for how to use them). >> >>Restart Medialibrary, and see your beautifull work ! >> >> >>b. Installation messages and labels >> >>The IzPack installer is already fully internationalized, so nothing to = do here=20 >>besides telling him to let the user choose spanish as the installation = >>language as well. We'll do it. >> >> >>c. "About" and documentation pages >> >>For now there is no internationalization support of those (english = only);=20 >>moreover translating it s a bigger and more tedious work than the = previous >>one. If you wish to do it nonetheless, tell us and we'll put in the technical=20 >>framework required, as well as provide you with the document sources = to be >>translated. >> >>Finally, send us the files, and we will be pleased to include it to = the >>incoming 0.8 version of Medialibrary. >> >> >> =20 >> |
From: MaDProph <mad...@la...> - 2005-11-23 10:32:26
|
La classe=85 je compl=E8terai et g=E9n=E9rerai les screenshots =E0 = convertir ou que tu m=92indiqueras=85 d=92ici l=E0 je regarde ledit bug =20 +a =20 M. =20 _____ =20 De : oli...@ac... [mailto:oli...@ac...]=20 Envoy=E9 : mardi 22 novembre 2005 16:34 =C0 : mar...@la... Objet : RE:=20 =20 Je te refais une doc facon Accenture, lisible et belle, concise et = facile =E0 lire, avec Open Office =20 Je te l'envoie =E0 la fin dde la semaine pour retours. =20 olivier =20 _____ =20 From: Marc Dutoo [mailto:mar...@la...]=20 Sent: 22 novembre 2005 15:37 To: Davy, Olivier Subject: RE:=20 ;_ ; laisse tomber les screenshots alors, le but =E9tait d'avoir une doc = digne de la gueule de la b=EAte. =20 Pour le format de la doc, celui qui permet de r=E9utiliser au maximum la = plus compl=E8te et =E0 jour doc qu'on ait aujourd'hui ! =20 Pour le dev, moi pas mieux : j'ai crad=E9 trois trucs en huit mois (hors = des bidouilles et du scripting) -_- =20 _____ =20 De : oli...@ac... [mailto:oli...@ac...]=20 Envoy=E9 : mardi 22 novembre 2005 15:23 =C0 : mar...@la... Objet : RE:=20 =20 Sur mon portable ACN, je n'ai que la JDK 1.4.2. =20 Je propose de rester la dessus dans un premier temps, car je ne vais pas pouvoir trop t=E9l=E9charger de choses dessus. =20 olive, qui n'a plus fait de DEV depuis plus de 5 mois. =20 =20 =20 _____ =20 From: Marc Dutoo [mailto:mar...@la...]=20 Sent: 22 novembre 2005 15:10 To: Davy, Olivier Subject: RE:=20 ^^ Que voil=E0 une heureuse pr=E9misse =20 A mon sens pour une 0.8 qui ne soit pas du renommage de fichier mais une release il reste * mettre la doc =E0 jour avec la nouvelle interface et les nouvelles fonctionnalit=E9s (instructions et captures d'=E9cran) !! La preuve, = l'ami Marek qui demande comment on cherche dans les commentaires * un bug =E0 v=E9rifier (recherche dans les commentaires selon Marek) et = une install =E0 tester * une news =E0 poster sur sf.net et surtout sur notre site (voire une = vraie campagne de comm, framasoft, rameuter les utilisateurs existants etc = etc) =20 Le plus gros, c'est la doc. On peut se le partager comme =E7a : * =E0 toi la mise =E0 jour de la doc qui est encore valable et les screenshots correspondants en jdk 1.5=20 * =E0 moi la doc des nouveaux trucs (recherche filtr=E9e etc.) et les screenshots correspondants=20 =20 Le reste je peux m'en charger, avec ton support (et celui des autres = rigolos <BAF> euh). On pourrait ainsi avoir une release ce week end. non ?? si !! =20 Qu'en dis-tu ? Je forwarde =E0 medialib-devel ? =20 _____ =20 De : oli...@ac... [mailto:oli...@ac...]=20 Envoy=E9 : mardi 22 novembre 2005 14:51 =C0 : mar...@fr... Objet :=20 =20 Bon, on passe ML en version 0.8 en enlevant la pre-release ? =20 olive =20 Olivier DAVY Accenture - Government Atlantic & Europe=20 Paris Office, France=20 Dial: +33 1 53 23 48 04=20 Email: oli...@ac... =20 This message is for the designated recipient only and may contain privileged, proprietary, or otherwise private information. If you have received it in error, please notify the sender immediately and delete = the original. Any other use of the email by you is prohibited. This message is for the designated recipient only and may contain privileged, proprietary, or otherwise private information. If you have received it in error, please notify the sender immediately and delete = the original. Any other use of the email by you is prohibited. This message is for the designated recipient only and may contain privileged, proprietary, or otherwise private information. If you have received it in error, please notify the sender immediately and delete = the original. Any other use of the email by you is prohibited. |
From: MaDProph <mad...@la...> - 2005-06-15 18:48:59
|
Hi Gon=E7alo =46irst thanks for your many thanks ! I'm really happy to know you like it.= =20 Moreover, we need user feedback such as yours to improve it further. Maybe I can help you right now. Font and size can't be changed for now; but= =20 there's a hidden feature that lets you change the icon set ! Trigger the me= nu=20 item Edit > Preferences ; then set the "medialib.icons.directory" to=20 "small_gui" instead of "glass"; then restart medialib. You'll see the icons= =20 are much smaller (but somewhat uglier) ! (you could even design your own ic= on=20 set). Thanks for the help offer. In fact we already have the portuguese translati= on=20 and it will be included in the 0.8 final release (still polishing=20 documentation). But I'll ask you if we lose it till then ;) Happy collecting M. Le mercredi 15 Juin 2005 01:11, vous avez =E9crit=A0: > Hello, > > I'm a media library user, and I really find it usefull! One thing that > I would like to be able to do is, change the FONT and the SIZE... possibly > change the icon that appears... get it smaller so more entries can be sho= wn > on screen. > > Please try to include it in the next version. It would be a great > feature! Keep up the excelent work, and allways in free software! If you > want I can translate the media library program to Portuguese. > > Gon=E7alo DeLemos F.M. > et...@ua... |
From: MaDProph <mad...@la...> - 2005-03-18 15:13:25
|
What about binary decorations ? are they allowed by the model ? (see mauricio's rfe on letting the user set an image for each medium) M. |
From: MaDProph <mad...@la...> - 2005-01-12 16:26:52
|
Le mercredi 12 Janvier 2005 16:12, Davy Olivier a =E9crit=A0: > Well, we should create the branch when we freeze the project to release i= t. > beta, Realease candidates, tests, should be conducted on the branch. > Thus Olive can still perform huge changes on the head in parallel ;) lol but remember ml2 development occurs in the ml2 cvs project, gui included M. > > > RCs 0.8 0.8.1 0.8.2 Few Release fixes (merge in > HEAD if +-+++--+--------+------+---------> 0.8 Branch (STABLE) > > ---+------+-------+-----+---+--+---+---> Trunk / HEAD (UNSTABLE) > 0.7.9 Freeze 0.9 0.9.1 | > +---------> 1.0 Branch (STABLE) > > > > -----Message d'origine----- > De : med...@li... > [mailto:med...@li...]De la part de > MaDProph > Envoy=E9 : mercredi 12 janvier 2005 16:07 > =C0 : med...@li... > Objet : Fwd: [Medialibrary-devel] Version management > > > > Already said, nice idea. > Let's create it when the actual release is done. > > M. > > ---------- Message transmis ---------- > > Subject: [Medialibrary-devel] Version management > Date: mercredi 12 Janvier 2005 13:42 > From: "Davy Olivier" <od...@so...> > To: <med...@li...> > > I would like to ask wheither or not we should use a branch strategy when > publishing stable release ? See the graph below. > It could enable our team to fix quickly bugs in release versions (well, s= ee > the bugs submitted, such bugs exist). > > > 0.8.1 0.8.2 Few Release fixes (merge in HEAD if > necessary) +----+-----------+---------> 0.8 Branch (STABLE) > > ---+------+-------+-----+---+--+---+---> Trunk / HEAD (UNSTABLE) > 0.7.9 0.9 0.9.1 | > +---------> 1.0 Branch (STABLE) > > > ------------------------------------------------------- > The SF.Net email is sponsored by: Beat the post-holiday blues > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt > _______________________________________________ > Medialibrary-devel mailing list > Med...@li... > https://lists.sourceforge.net/lists/listinfo/medialibrary-devel > > ------------------------------------------------------- > > > ------------------------------------------------------- > The SF.Net email is sponsored by: Beat the post-holiday blues > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt > _______________________________________________ > Medialibrary-devel mailing list > Med...@li... > https://lists.sourceforge.net/lists/listinfo/medialibrary-devel > > > ------------------------------------------------------- > The SF.Net email is sponsored by: Beat the post-holiday blues > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt > _______________________________________________ > Medialibrary-devel mailing list > Med...@li... > https://lists.sourceforge.net/lists/listinfo/medialibrary-devel |
From: Davy O. <od...@so...> - 2005-01-12 15:13:00
|
Well, we should create the branch when we freeze the project to release = it. beta, Realease candidates, tests, should be conducted on the branch. Thus Olive can still perform huge changes on the head in parallel ;) RCs 0.8 0.8.1 0.8.2 Few Release fixes (merge in = HEAD if +-+++--+--------+------+---------> 0.8 Branch (STABLE) | ---+------+-------+-----+---+--+---+---> Trunk / HEAD (UNSTABLE) 0.7.9 Freeze 0.9 0.9.1 | +---------> 1.0 Branch (STABLE) -----Message d'origine----- De : med...@li... [mailto:med...@li...]De la part de MaDProph Envoy=E9 : mercredi 12 janvier 2005 16:07 =C0 : med...@li... Objet : Fwd: [Medialibrary-devel] Version management Already said, nice idea. Let's create it when the actual release is done. M. ---------- Message transmis ---------- Subject: [Medialibrary-devel] Version management Date: mercredi 12 Janvier 2005 13:42 From: "Davy Olivier" <od...@so...> To: <med...@li...> I would like to ask wheither or not we should use a branch strategy when publishing stable release ? See the graph below. It could enable our team to fix quickly bugs in release versions (well, = see the bugs submitted, such bugs exist). 0.8.1 0.8.2 Few Release fixes (merge in HEAD = if necessary) +----+-----------+---------> 0.8 Branch (STABLE) ---+------+-------+-----+---+--+---+---> Trunk / HEAD (UNSTABLE) 0.7.9 0.9 0.9.1 | +---------> 1.0 Branch (STABLE) ------------------------------------------------------- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt _______________________________________________ Medialibrary-devel mailing list Med...@li... https://lists.sourceforge.net/lists/listinfo/medialibrary-devel ------------------------------------------------------- ------------------------------------------------------- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt _______________________________________________ Medialibrary-devel mailing list Med...@li... https://lists.sourceforge.net/lists/listinfo/medialibrary-devel |
From: MaDProph <mad...@la...> - 2005-01-12 15:02:28
|
Already said, nice idea. Let's create it when the actual release is done. M. ---------- Message transmis ---------- Subject: [Medialibrary-devel] Version management Date: mercredi 12 Janvier 2005 13:42 From: "Davy Olivier" <od...@so...> To: <med...@li...> I would like to ask wheither or not we should use a branch strategy when publishing stable release ? See the graph below. It could enable our team to fix quickly bugs in release versions (well, see the bugs submitted, such bugs exist). 0.8.1 0.8.2 Few Release fixes (merge in HEAD if necessary) +----+-----------+---------> 0.8 Branch (STABLE) ---+------+-------+-----+---+--+---+---> Trunk / HEAD (UNSTABLE) 0.7.9 0.9 0.9.1 | +---------> 1.0 Branch (STABLE) ------------------------------------------------------- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt _______________________________________________ Medialibrary-devel mailing list Med...@li... https://lists.sourceforge.net/lists/listinfo/medialibrary-devel ------------------------------------------------------- |
From: Davy O. <od...@so...> - 2005-01-12 12:42:53
|
I would like to ask wheither or not we should use a branch strategy when = publishing stable release ? See the graph below. It could enable our team to fix quickly bugs in release versions (well, = see the bugs submitted, such bugs exist). 0.8.1 0.8.2 Few Release fixes (merge in HEAD = if necessary) +----+-----------+---------> 0.8 Branch (STABLE) | ---+------+-------+-----+---+--+---+---> Trunk / HEAD (UNSTABLE) 0.7.9 0.9 0.9.1 | +---------> 1.0 Branch (STABLE) |
From: Davy O. <od...@so...> - 2005-01-07 08:41:30
|
Seen on linufr : Les fichiers d'OpenOffice.org sont en XML. DocBook est un standard XML = OASIS de documentation technique. Plusieurs approches sont possibles pour passer de l'un =E0 l'autre. = Indesko a poursuivi le travail d'Eric Bellot (avec son accord) et met = ses =E9volutions =E0 disposition de la communaut=E9. OOo2DBK permet de cr=E9er des documents DocBook au format "article" et = "book". Outre les formatages classiques (comme emphasis), l'export = g=E8re bibliographie, glossaire, pr=E9face, annexes, table des = mati=E8res, images ins=E9r=E9es dans le document (et redimensionnement), = metadata =E0 l'aide de champs utilisateurs, bordures de tableaux etc ... OpenOffice.org, devient ainsi, un =E9diteur graphique simple =E0 prendre = en main pour produire du DocBook XML. Indesko esp=E8re par cette initiative mettre =E0 la port=E9e de tous la = production de documents DocBook, leur permettant alors de b=E9n=E9ficier = de tous les outils de traitement DocBook. NdMeR : le format DocBook (que ce soit en XML ou SGML) est un format = normalis=E9 par OASIS Open, la m=EAme organisation qui est en cours de = normalisation du format XML zipp=E9 utilis=E9 par la suite = OpenOffice.org/StarOffice et prochainement KOffice, la suite bureautique = de KDE. * [fr] OOo2DBK sur le site d'Indesko (309 hits) * [en] OASIS Open (48 hits) |
From: MaDProph <mad...@la...> - 2005-01-06 17:11:39
|
Those numbers will have great value for us. Thx for gathering em Olive ! M. Le jeudi 6 Janvier 2005 17:43, Davy Olivier a =E9crit=A0: > The last statistics were not correct due to the File <--> Media Associati= on > problem. > > The import is now correct (I have checked the number against ML1) : > > mysql> select count(*) from file where directory like "0"; > +----------+ > > | count(*) | > > +----------+ > > | 205426 | > > +----------+ > 1 row in set (1.69 sec) > > mysql> select count(*) from file where directory like "1"; > +----------+ > > | count(*) | > > +----------+ > > | 14016 | > > +----------+ > 1 row in set (1.66 sec) > > The count(*) queries go faster after having optimize the file table (than= ks > to Nunux) : optimize table file; > or rebooting mysql. > > > Olive > > > ------------------------------------------------------- > The SF.Net email is sponsored by: Beat the post-holiday blues > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt > _______________________________________________ > Medialibrary-devel mailing list > Med...@li... > https://lists.sourceforge.net/lists/listinfo/medialibrary-devel |
From: Davy O. <od...@so...> - 2005-01-06 16:43:47
|
The last statistics were not correct due to the File <--> Media = Association problem. The import is now correct (I have checked the number against ML1) : mysql> select count(*) from file where directory like "0"; +----------+ | count(*) | +----------+ | 205426 | +----------+ 1 row in set (1.69 sec) mysql> select count(*) from file where directory like "1"; +----------+ | count(*) | +----------+ | 14016 | +----------+ 1 row in set (1.66 sec) The count(*) queries go faster after having optimize the file table = (thanks to Nunux) : optimize table file; or rebooting mysql. Olive |
From: Davy O. <od...@so...> - 2005-01-06 15:41:10
|
I got a problem with the model File -- Media=20 In the parent File, when adding a subfile named file : File (file) <--> Parent (this) <--> Media (this.media) this.media.add(file); is incorrect. It provokes a = TransientObjectException when inserterd Since it adds File to the Set files, File (file) <--> Media (this.media) when we flush the Media, Hibernate could flush the File before its = Parent. But the File.getParent() is transient (not already flush), and it leads = to a TransientObjectException So it is recommended that only root files are added in the = Media.getFiles() Set. Then getFiles() is not implemented by a HashSet (files should remain = null), but by a hibernate query. Only rootFiles should be a HashSet, and it is through rootFiles that the = media will be flushed. And then recursively, the parents will flush the files. Consequences :=20 - we must store in media only a Set of rootFiles, no all the files. - we should delete media.getFiles() from the bean and only keep = media.getRootFiles(); - If one wants to access to all the files, then we will create a method = Navigation.getAllFiles(media), which execute the right SQL query (fast, = since we have the media id in each file). olive |
From: Davy O. <od...@so...> - 2005-01-05 16:50:28
|
Some stats from a "Mad mysql" catalog files : 258477 dont 52035 directories medias : 1329 zip files : 3394 mysql> select count(*) from file where (name like "%love%") and (name = like "%hi a%"); +----------+ | count(*) | +----------+ | 91 | +----------+ 1 row in set (14.81 sec) WHY ? ---- select count(*) takes 14s. select * takes 2s. FOR FUN (it is so easy no to compute stats)=20 ------------------------------------------- mysql> select count(*) from file where name like "%.mp3"; +----------+ | count(*) | +----------+ | 5105 | +----------+ 1 row in set (14.75 sec) mysql> select sum(size) from file where name like "%.mp3"; +-------------+ | sum(size) | +-------------+ | 23170215115 | +-------------+ 1 row in set (2.14 sec) mysql> select count(*) from file where name like "%.avi"; +----------+ | count(*) | +----------+ | 5204 | +----------+ 1 row in set (14.75 sec) mysql> select sum(size) from file where name like "%.avi"; +--------------+ | sum(size) | +--------------+ | 667369113832 | +--------------+ 1 row in set (2.14 sec) mysql> select avg(size) from file where name like "%.avi"; +----------------+ | avg(size) | +----------------+ | 128241566.8394 | +----------------+ 1 row in set (2.14 sec) ohoho 128 !!! The bit Mystery ! olive |
From: Davy O. <od...@so...> - 2005-01-05 16:32:29
|
well, a good architecture leads to freedom of choice ;) Packaging mysql is not so hard (I got the script for a standalone = execution, it is very easy). The search and stats with MySQL may be faster than in ML1. Hope So ! olive -----Message d'origine----- De : med...@li... [mailto:med...@li...]De la part de MaDProph Envoy=E9 : mercredi 5 janvier 2005 17:21 =C0 : med...@li... Objet : Re: [Medialibrary-devel] ML2: MySQL vs HSQLDB performances good tests olive I agree, we should be flexible on the DB choice, at least at integration = level. we could even make one release with hsqldb and one with mysql. I guess the difficulty would be in nicely packaging mysql (in an = installer,=20 with a good default configuration etc. etc.). M. Le mercredi 5 Janvier 2005 16:51, Davy Olivier a =E9crit=A0: > Here are performance tests > > > import of Mad cat > MySQL innoDB (for transactions) on localhost =3D 470s > HSQLDB =3D 290s > > search "java doc" (and other different search) > MySQL =3D 2s > HSQLDB =3D 10s > > Conclusion : the import (insert querie) is longer for mysql, since it = use a > pseudo network connection but the search (and even the other select or > count(*)) are of course faster on MysqlDB, which is a true DBServer, = with > great indexes, etc. > > Let the user choose one of the solution for his catalog : Standalone = HSQLDB > vs Server MySQL (personnaly, since I am an engineer, I will use mysql = for > my catalog, faster and more secure). > > olive > P.S. I will commit into CVS the mysql database creation soon. > > -----Message d'origine----- > De : med...@li... > [mailto:med...@li...]De la part de > MaDProph > Envoy=E9 : mercredi 5 janvier 2005 14:55 > =C0 : med...@li... > Objet : Re: [Medialibrary-devel] Re: ML 2 search > > > > It should be possible in SQL : create index plouf on table(champ); but = the > way u're doing it (through hibernate ?), is it still possible ? or = even in > jdbc ? > > olive : about the tag & branch thingy, I did actually agree with u if = you > look at the end of my mail (just don't branch for unstable releases, > obviously). > > M. > > Le mercredi 5 Janvier 2005 14:22, Davy Olivier a =E9crit=A0: > > Thanks for the tip. > > > > Well, that is true, I know that optimization. > > But is it possible with HSQLDB (i am a newbie in hsqldb)? I have to = find > > it out ! And is it necessary for 20 seconds of an import that is = done > > only once... > > > > I will try to make a comparison with MySQL soon (import + search = perf). > > > > Olive > > > > -----Message d'origine----- > > De : med...@li... > > [mailto:med...@li...]De la part de > > Vincent KONIECZEK > > Envoy=E9 : mercredi 5 janvier 2005 12:42 > > =C0 : mad...@la... > > Cc : med...@li... > > Objet : Re: [Medialibrary-devel] Re: ML 2 search > > > > > > I do not understand exactly the difference between the two versions = but > > a rule in database datamining when importing data is to disable the > > index mechanism during the import and to create the index once the > > import is finished. > > This enables better performance since there is no consistency checks = nor > > tree computing to make. > > > > > > V > > > > >Answering to both of your mails : > > > > > >we're bound to have worse performance than in ML1 since, as u say, = ML1 > > > keeps everything in memory. So I think this kind of search = performance > > > is ok. Obviously import time is not a problem as long as its = purpose is > > > optimizing the main, day-to-day uses of ML2. > > > > > >I have no definitive answer on import yet. We'll have to think a = little > > > bit more of the user experience of this feature. > > > > > >M. > > > > > >Le mercredi 5 Janvier 2005 11:42, vous avez =E9crit : > > >>je passe de 13 =E0 10 secondes pour la recherche de "java doc" sur = ton > > >>catalogue avec un index sur le file name. Sur ML 1, elle est de 5 > > >> secondes (normal, tout le catalaogue est en m=E9moire, soit = 110MB). > > >> > > >>L'index fait passer de 263 =E0 293s l'import de catalogue (soit = 10% de > > >> plus). > > >> > > >> > > >>o. > > > > > >------------------------------------------------------- > > >The SF.Net email is sponsored by: Beat the post-holiday blues > > >Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. > > >It's fun and FREE -- well, = almost....http://www.thinkgeek.com/sfshirt > > >_______________________________________________ > > >Medialibrary-devel mailing list > > >Med...@li... > > >https://lists.sourceforge.net/lists/listinfo/medialibrary-devel > > > > ------------------------------------------------------- > > The SF.Net email is sponsored by: Beat the post-holiday blues > > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. > > It's fun and FREE -- well, = almost....http://www.thinkgeek.com/sfshirt > > _______________________________________________ > > Medialibrary-devel mailing list > > Med...@li... > > https://lists.sourceforge.net/lists/listinfo/medialibrary-devel > > ------------------------------------------------------- > The SF.Net email is sponsored by: Beat the post-holiday blues > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt > _______________________________________________ > Medialibrary-devel mailing list > Med...@li... > https://lists.sourceforge.net/lists/listinfo/medialibrary-devel > > > ------------------------------------------------------- > The SF.Net email is sponsored by: Beat the post-holiday blues > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt > _______________________________________________ > Medialibrary-devel mailing list > Med...@li... > https://lists.sourceforge.net/lists/listinfo/medialibrary-devel ------------------------------------------------------- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt _______________________________________________ Medialibrary-devel mailing list Med...@li... https://lists.sourceforge.net/lists/listinfo/medialibrary-devel |
From: MaDProph <mad...@la...> - 2005-01-05 16:17:39
|
good tests olive I agree, we should be flexible on the DB choice, at least at integration=20 level. we could even make one release with hsqldb and one with mysql. I guess the difficulty would be in nicely packaging mysql (in an installer,= =20 with a good default configuration etc. etc.). M. Le mercredi 5 Janvier 2005 16:51, Davy Olivier a =E9crit=A0: > Here are performance tests > > > import of Mad cat > MySQL innoDB (for transactions) on localhost =3D 470s > HSQLDB =3D 290s > > search "java doc" (and other different search) > MySQL =3D 2s > HSQLDB =3D 10s > > Conclusion : the import (insert querie) is longer for mysql, since it use= a > pseudo network connection but the search (and even the other select or > count(*)) are of course faster on MysqlDB, which is a true DBServer, with > great indexes, etc. > > Let the user choose one of the solution for his catalog : Standalone HSQL= DB > vs Server MySQL (personnaly, since I am an engineer, I will use mysql for > my catalog, faster and more secure). > > olive > P.S. I will commit into CVS the mysql database creation soon. > > -----Message d'origine----- > De : med...@li... > [mailto:med...@li...]De la part de > MaDProph > Envoy=E9 : mercredi 5 janvier 2005 14:55 > =C0 : med...@li... > Objet : Re: [Medialibrary-devel] Re: ML 2 search > > > > It should be possible in SQL : create index plouf on table(champ); but the > way u're doing it (through hibernate ?), is it still possible ? or even in > jdbc ? > > olive : about the tag & branch thingy, I did actually agree with u if you > look at the end of my mail (just don't branch for unstable releases, > obviously). > > M. > > Le mercredi 5 Janvier 2005 14:22, Davy Olivier a =E9crit=A0: > > Thanks for the tip. > > > > Well, that is true, I know that optimization. > > But is it possible with HSQLDB (i am a newbie in hsqldb)? I have to find > > it out ! And is it necessary for 20 seconds of an import that is done > > only once... > > > > I will try to make a comparison with MySQL soon (import + search perf). > > > > Olive > > > > -----Message d'origine----- > > De : med...@li... > > [mailto:med...@li...]De la part de > > Vincent KONIECZEK > > Envoy=E9 : mercredi 5 janvier 2005 12:42 > > =C0 : mad...@la... > > Cc : med...@li... > > Objet : Re: [Medialibrary-devel] Re: ML 2 search > > > > > > I do not understand exactly the difference between the two versions but > > a rule in database datamining when importing data is to disable the > > index mechanism during the import and to create the index once the > > import is finished. > > This enables better performance since there is no consistency checks nor > > tree computing to make. > > > > > > V > > > > >Answering to both of your mails : > > > > > >we're bound to have worse performance than in ML1 since, as u say, ML1 > > > keeps everything in memory. So I think this kind of search performance > > > is ok. Obviously import time is not a problem as long as its purpose = is > > > optimizing the main, day-to-day uses of ML2. > > > > > >I have no definitive answer on import yet. We'll have to think a little > > > bit more of the user experience of this feature. > > > > > >M. > > > > > >Le mercredi 5 Janvier 2005 11:42, vous avez =E9crit : > > >>je passe de 13 =E0 10 secondes pour la recherche de "java doc" sur ton > > >>catalogue avec un index sur le file name. Sur ML 1, elle est de 5 > > >> secondes (normal, tout le catalaogue est en m=E9moire, soit 110MB). > > >> > > >>L'index fait passer de 263 =E0 293s l'import de catalogue (soit 10% de > > >> plus). > > >> > > >> > > >>o. > > > > > >------------------------------------------------------- > > >The SF.Net email is sponsored by: Beat the post-holiday blues > > >Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. > > >It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt > > >_______________________________________________ > > >Medialibrary-devel mailing list > > >Med...@li... > > >https://lists.sourceforge.net/lists/listinfo/medialibrary-devel > > > > ------------------------------------------------------- > > The SF.Net email is sponsored by: Beat the post-holiday blues > > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. > > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt > > _______________________________________________ > > Medialibrary-devel mailing list > > Med...@li... > > https://lists.sourceforge.net/lists/listinfo/medialibrary-devel > > ------------------------------------------------------- > The SF.Net email is sponsored by: Beat the post-holiday blues > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt > _______________________________________________ > Medialibrary-devel mailing list > Med...@li... > https://lists.sourceforge.net/lists/listinfo/medialibrary-devel > > > ------------------------------------------------------- > The SF.Net email is sponsored by: Beat the post-holiday blues > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt > _______________________________________________ > Medialibrary-devel mailing list > Med...@li... > https://lists.sourceforge.net/lists/listinfo/medialibrary-devel |
From: Davy O. <od...@so...> - 2005-01-05 15:51:49
|
Here are performance tests import of Mad cat MySQL innoDB (for transactions) on localhost =3D 470s HSQLDB =3D 290s search "java doc" (and other different search) MySQL =3D 2s HSQLDB =3D 10s Conclusion : the import (insert querie) is longer for mysql, since it = use a pseudo network connection but the search (and even the other select or count(*)) are of course = faster on MysqlDB, which is a true DBServer, with great indexes, etc. Let the user choose one of the solution for his catalog : Standalone = HSQLDB vs Server MySQL (personnaly, since I am an engineer, I will use mysql for my catalog, = faster and more secure). olive P.S. I will commit into CVS the mysql database creation soon. -----Message d'origine----- De : med...@li... [mailto:med...@li...]De la part de MaDProph Envoy=E9 : mercredi 5 janvier 2005 14:55 =C0 : med...@li... Objet : Re: [Medialibrary-devel] Re: ML 2 search It should be possible in SQL : create index plouf on table(champ); but = the way=20 u're doing it (through hibernate ?), is it still possible ? or even in = jdbc ? olive : about the tag & branch thingy, I did actually agree with u if = you look=20 at the end of my mail (just don't branch for unstable releases, = obviously). M. Le mercredi 5 Janvier 2005 14:22, Davy Olivier a =E9crit=A0: > Thanks for the tip. > > Well, that is true, I know that optimization. > But is it possible with HSQLDB (i am a newbie in hsqldb)? I have to = find it > out ! And is it necessary for 20 seconds of an import that is done = only > once... > > I will try to make a comparison with MySQL soon (import + search = perf). > > Olive > > -----Message d'origine----- > De : med...@li... > [mailto:med...@li...]De la part de > Vincent KONIECZEK > Envoy=E9 : mercredi 5 janvier 2005 12:42 > =C0 : mad...@la... > Cc : med...@li... > Objet : Re: [Medialibrary-devel] Re: ML 2 search > > > I do not understand exactly the difference between the two versions = but > a rule in database datamining when importing data is to disable the > index mechanism during the import and to create the index once the > import is finished. > This enables better performance since there is no consistency checks = nor > tree computing to make. > > > V > > >Answering to both of your mails : > > > >we're bound to have worse performance than in ML1 since, as u say, = ML1 > > keeps everything in memory. So I think this kind of search = performance is > > ok. Obviously import time is not a problem as long as its purpose is > > optimizing the main, day-to-day uses of ML2. > > > >I have no definitive answer on import yet. We'll have to think a = little > > bit more of the user experience of this feature. > > > >M. > > > >Le mercredi 5 Janvier 2005 11:42, vous avez =E9crit : > >>je passe de 13 =E0 10 secondes pour la recherche de "java doc" sur = ton > >>catalogue avec un index sur le file name. Sur ML 1, elle est de 5 > >> secondes (normal, tout le catalaogue est en m=E9moire, soit 110MB). > >> > >>L'index fait passer de 263 =E0 293s l'import de catalogue (soit 10% = de > >> plus). > >> > >> > >>o. > > > >------------------------------------------------------- > >The SF.Net email is sponsored by: Beat the post-holiday blues > >Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. > >It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt > >_______________________________________________ > >Medialibrary-devel mailing list > >Med...@li... > >https://lists.sourceforge.net/lists/listinfo/medialibrary-devel > > ------------------------------------------------------- > The SF.Net email is sponsored by: Beat the post-holiday blues > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt > _______________________________________________ > Medialibrary-devel mailing list > Med...@li... > https://lists.sourceforge.net/lists/listinfo/medialibrary-devel ------------------------------------------------------- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt _______________________________________________ Medialibrary-devel mailing list Med...@li... https://lists.sourceforge.net/lists/listinfo/medialibrary-devel |
From: MaDProph <mad...@la...> - 2005-01-05 13:51:02
|
It should be possible in SQL : create index plouf on table(champ); but the = way=20 u're doing it (through hibernate ?), is it still possible ? or even in jdbc= ? olive : about the tag & branch thingy, I did actually agree with u if you l= ook=20 at the end of my mail (just don't branch for unstable releases, obviously). M. Le mercredi 5 Janvier 2005 14:22, Davy Olivier a =E9crit=A0: > Thanks for the tip. > > Well, that is true, I know that optimization. > But is it possible with HSQLDB (i am a newbie in hsqldb)? I have to find = it > out ! And is it necessary for 20 seconds of an import that is done only > once... > > I will try to make a comparison with MySQL soon (import + search perf). > > Olive > > -----Message d'origine----- > De : med...@li... > [mailto:med...@li...]De la part de > Vincent KONIECZEK > Envoy=E9 : mercredi 5 janvier 2005 12:42 > =C0 : mad...@la... > Cc : med...@li... > Objet : Re: [Medialibrary-devel] Re: ML 2 search > > > I do not understand exactly the difference between the two versions but > a rule in database datamining when importing data is to disable the > index mechanism during the import and to create the index once the > import is finished. > This enables better performance since there is no consistency checks nor > tree computing to make. > > > V > > >Answering to both of your mails : > > > >we're bound to have worse performance than in ML1 since, as u say, ML1 > > keeps everything in memory. So I think this kind of search performance = is > > ok. Obviously import time is not a problem as long as its purpose is > > optimizing the main, day-to-day uses of ML2. > > > >I have no definitive answer on import yet. We'll have to think a little > > bit more of the user experience of this feature. > > > >M. > > > >Le mercredi 5 Janvier 2005 11:42, vous avez =E9crit : > >>je passe de 13 =E0 10 secondes pour la recherche de "java doc" sur ton > >>catalogue avec un index sur le file name. Sur ML 1, elle est de 5 > >> secondes (normal, tout le catalaogue est en m=E9moire, soit 110MB). > >> > >>L'index fait passer de 263 =E0 293s l'import de catalogue (soit 10% de > >> plus). > >> > >> > >>o. > > > >------------------------------------------------------- > >The SF.Net email is sponsored by: Beat the post-holiday blues > >Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. > >It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt > >_______________________________________________ > >Medialibrary-devel mailing list > >Med...@li... > >https://lists.sourceforge.net/lists/listinfo/medialibrary-devel > > ------------------------------------------------------- > The SF.Net email is sponsored by: Beat the post-holiday blues > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt > _______________________________________________ > Medialibrary-devel mailing list > Med...@li... > https://lists.sourceforge.net/lists/listinfo/medialibrary-devel |
From: Davy O. <od...@so...> - 2005-01-05 13:22:58
|
Thanks for the tip. Well, that is true, I know that optimization. But is it possible with HSQLDB (i am a newbie in hsqldb)? I have to find = it out ! And is it necessary for 20 seconds of an import that is done only = once... I will try to make a comparison with MySQL soon (import + search perf). Olive -----Message d'origine----- De : med...@li... [mailto:med...@li...]De la part de Vincent KONIECZEK Envoy=E9 : mercredi 5 janvier 2005 12:42 =C0 : mad...@la... Cc : med...@li... Objet : Re: [Medialibrary-devel] Re: ML 2 search I do not understand exactly the difference between the two versions but=20 a rule in database datamining when importing data is to disable the=20 index mechanism during the import and to create the index once the=20 import is finished. This enables better performance since there is no consistency checks nor = tree computing to make. V >Answering to both of your mails : > >we're bound to have worse performance than in ML1 since, as u say, ML1 = keeps=20 >everything in memory. So I think this kind of search performance is ok. = >Obviously import time is not a problem as long as its purpose is = optimizing=20 >the main, day-to-day uses of ML2. > >I have no definitive answer on import yet. We'll have to think a little = bit=20 >more of the user experience of this feature. > >M. > >Le mercredi 5 Janvier 2005 11:42, vous avez =E9crit : > =20 > >>je passe de 13 =E0 10 secondes pour la recherche de "java doc" sur ton >>catalogue avec un index sur le file name. Sur ML 1, elle est de 5 = secondes >>(normal, tout le catalaogue est en m=E9moire, soit 110MB). >> >>L'index fait passer de 263 =E0 293s l'import de catalogue (soit 10% de = plus). >> >> >>o. >> =20 >> > > >------------------------------------------------------- >The SF.Net email is sponsored by: Beat the post-holiday blues >Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. >It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt >_______________________________________________ >Medialibrary-devel mailing list >Med...@li... >https://lists.sourceforge.net/lists/listinfo/medialibrary-devel > > =20 > |
From: Vincent K. <vin...@st...> - 2005-01-05 13:12:32
|
I do not understand exactly the difference between the two versions but=20 a rule in database datamining when importing data is to disable the=20 index mechanism during the import and to create the index once the=20 import is finished. This enables better performance since there is no consistency checks nor=20 tree computing to make. V >Answering to both of your mails : > >we're bound to have worse performance than in ML1 since, as u say, ML1 k= eeps=20 >everything in memory. So I think this kind of search performance is ok.=20 >Obviously import time is not a problem as long as its purpose is optimiz= ing=20 >the main, day-to-day uses of ML2. > >I have no definitive answer on import yet. We'll have to think a little = bit=20 >more of the user experience of this feature. > >M. > >Le mercredi 5 Janvier 2005 11:42, vous avez =E9crit : > =20 > >>je passe de 13 =E0 10 secondes pour la recherche de "java doc" sur ton >>catalogue avec un index sur le file name. Sur ML 1, elle est de 5 secon= des >>(normal, tout le catalaogue est en m=E9moire, soit 110MB). >> >>L'index fait passer de 263 =E0 293s l'import de catalogue (soit 10% de = plus). >> >> >>o. >> =20 >> > > >------------------------------------------------------- >The SF.Net email is sponsored by: Beat the post-holiday blues >Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. >It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt >_______________________________________________ >Medialibrary-devel mailing list >Med...@li... >https://lists.sourceforge.net/lists/listinfo/medialibrary-devel > > =20 > |
From: Davy O. <od...@so...> - 2005-01-05 12:36:08
|
Quoting Marc, " I agree, In a perfect world it'd be to way to go. The problem with branching is that backporting fixes through merges is = still a=20 tedious, time consuming task; and time we alas don't have much. " Well, there should a no or few little bugs in branch releases. So = backporting fixes from release branch to head is not really time = consuming, according to me. What would have been time consuming is if we decided, as it is done in = freebsd, to create stable branches... But we will not do that. Don't you think so ? olive |
From: MaDProph <mad...@la...> - 2005-01-05 12:11:42
|
I agree, In a perfect world it'd be to way to go. The problem with branching is that backporting fixes through merges is stil= l a=20 tedious, time consuming task; and time we alas don't have much. However, if we want to step up to a broader audience, and for the sake of=20 giving some true meaning to the "stable" concept, we should at least give i= t=20 a try. So I'd be ok for ur scheme. NB. remember "published in jar" means a jar installer. Binary installers ar= e=20 require some more work (either with InstallAnywhere or with the optional=20 IzPack module). And yep, I forgot to tag the two previous releases. Bad, bad me. For bug pa= tch=20 purpose, it's still possible to checkout ML at the date of the release and= =20 get the exact source of the release (using the CVS command line, not=20 eclipse). M. Le mercredi 5 Janvier 2005 12:17, Davy Olivier a =E9crit=A0: > Why not having 3 branches (and kind of files on sourceforge) of ML : > - unstable / current (0.7.8), tagged from CVS, TAG =3D ML_UNSTABLE_0_7_8, > never published on sf files. - stable / testing (0.7.9, far stable enough= ), > published in jar, TAG =3D ML_STABLE_0_7_9 - release (0.6, 0.8), installer, > BRANCH & TAG =3D ML_STABLE_0_6, ML_RELEASE_0_6_1 > > We should tag more often the CVS tree. > > We could create "secutity fixes" for releases (as for 0.6 that contains > important problems, and that is still tried by some users as far as I > know). To do that, we should create a release branch for each release, to > be able to develop fixes on that branch. > > Could be interesting to try that. > Then we will publish our work more often ! > > olive > > > ------------------------------------------------------- > The SF.Net email is sponsored by: Beat the post-holiday blues > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt > _______________________________________________ > Medialibrary-devel mailing list > Med...@li... > https://lists.sourceforge.net/lists/listinfo/medialibrary-devel |
From: Davy O. <od...@so...> - 2005-01-05 11:17:42
|
Why not having 3 branches (and kind of files on sourceforge) of ML :=20 - unstable / current (0.7.8), tagged from CVS, TAG =3D = ML_UNSTABLE_0_7_8, never published on sf files. - stable / testing (0.7.9, far stable enough), published in jar, TAG =3D = ML_STABLE_0_7_9 - release (0.6, 0.8), installer, BRANCH & TAG =3D ML_STABLE_0_6, = ML_RELEASE_0_6_1 We should tag more often the CVS tree. We could create "secutity fixes" for releases (as for 0.6 that contains = important problems, and that is still tried by some users as far as I = know). To do that, we should create a release branch for each release, = to be able to develop fixes on that branch. Could be interesting to try that. Then we will publish our work more often ! olive |
From: MaDProph <mad...@la...> - 2005-01-05 10:55:50
|
Answering to both of your mails : we're bound to have worse performance than in ML1 since, as u say, ML1 keep= s=20 everything in memory. So I think this kind of search performance is ok.=20 Obviously import time is not a problem as long as its purpose is optimizing= =20 the main, day-to-day uses of ML2. I have no definitive answer on import yet. We'll have to think a little bit= =20 more of the user experience of this feature. M. Le mercredi 5 Janvier 2005 11:42, vous avez =E9crit=A0: > je passe de 13 =E0 10 secondes pour la recherche de "java doc" sur ton > catalogue avec un index sur le file name. Sur ML 1, elle est de 5 secondes > (normal, tout le catalaogue est en m=E9moire, soit 110MB). > > L'index fait passer de 263 =E0 293s l'import de catalogue (soit 10% de pl= us). > > > o. |
From: MaDProph <mad...@la...> - 2005-01-03 10:33:13
|
Hi Olive Well done ! Let's keep up the good work. btw there'll be a binary installer for 0.8 (new optional IzPack module). Best of wishes for 2k5. M. > Le lundi 3 Janvier 2005 11:08, Davy Olivier a =E9crit=A0: > > Hi ! > > > > For the new year, I am please to announce that the second engine of > > MediaLibrary has been updateed : - improved JUnit tests, using a test > > catalog in /test/catalog ; > > - improvements in the model ; > > - XZCat catalog import business method implemented and fully tested > > (functionality, performances, memory consumption). No import of MP3 and > > Zip files for the moment). > > > > Cheers, > > > > Olive |