mt-daapd-devel Mailing List for Multi-Threaded DAAP Daemon (Page 2)
Status: Beta
Brought to you by:
andrew40
You can subscribe to this list here.
2005 |
Jan
|
Feb
(8) |
Mar
(55) |
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
(9) |
Oct
|
Nov
|
Dec
|
---|---|---|---|---|---|---|---|---|---|---|---|---|
2006 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
(2) |
Jul
|
Aug
|
Sep
|
Oct
|
Nov
(1) |
Dec
|
2007 |
Jan
(1) |
Feb
|
Mar
|
Apr
|
May
(2) |
Jun
|
Jul
(3) |
Aug
(1) |
Sep
|
Oct
|
Nov
|
Dec
|
2008 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
(1) |
Aug
|
Sep
|
Oct
(1) |
Nov
|
Dec
(1) |
2009 |
Jan
(1) |
Feb
|
Mar
(1) |
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
From: Ron P. <ro...@pe...> - 2005-03-18 00:28:26
|
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 > When I query something like this: > /databases/1/items?query='daap.songartist:Beatles','daap.album:da > greeeat album'&output=xml > I get a query with "or" between the artist and album. How do I make it > "and" /databases/1/items?query='daap.songartist:Beatles'%20'daap.album: da.greeeat album' Either a space or a + does AND. I noticed there is a problem with wildcard searches though. I'll take a look at that. -- Ron -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.3 (Darwin) iD8DBQFCOiClDl/VbZJe82cRAnAEAJ9wX9BbY24okrJ+0saptPqZJXuOJgCfTRUq 14F96IGH9IUfxbKQKdXzs7c= =kPFu -----END PGP SIGNATURE----- |
From: Ron P. <ro...@pe...> - 2005-03-18 00:14:42
|
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 > So this is what you can do on the "Playlist" tab in the web interface.=0D= > Click on "Library" will load your whole library and show it as a = table.=0D > If you click on the genre/artist/album boxes they will update and the =0D= > song list will reflect the selection.=0D =0D Cool...=0D =0D > databases/1/browse should return the results alphabetically (I guess =0D= > it is just a matter of adding =E2=80=9Corder by=E2=80=9D to the select = clauses.) (not =0D > a question I know)=0D =0D True... it returns them in whatever order it gets them in. Yeah, we =0D could choose how to order those. I might fiddle with that some this =0D weekend.=0D =0D > I can=E2=80=99t query some thing like this 'daap.songgenre:Children's = Music' I =0D > encode the ' with %27=0D =0D Oops. I'll fix that, too. Actually, I thought that was working, but =0D= obviously not.=0D =0D > Itunes sorts artists that begin with "the" on the name e.g. =E2=80=9CThe= =0D > Corrs=E2=80=9D will be under C not T. Should we do the same?=0D =0D We could do that, but I'm not sure the best way to do that. Maybe add =0D= another field that's the "sortable name", and always use that for =0D browses.=0D =0D > Can I through the daap queries tell the server how I want the result =0D= > sorted? "Gimme all Jackson songs sorted by song length"=0D =0D Not with default daap queries. But when I play with the other stuff, I =0D= don't see why we couldn't add something like a meta tag. =0D sort=3Ddaap.songartist,daap.songalbum=0D =0D to sort by artist first, then album.=0D =0D -- Ron=0D =0D -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.3 (Darwin) iD8DBQFCOh1uDl/VbZJe82cRAm6+AJ4nPjQvnTO88+q4ibf0qtPIpK1MmACgl2kA Gr99QYrnNG6hsoJJ0pJT+d4=3D =3Dcu+8 -----END PGP SIGNATURE----- |
From: <and...@us...> - 2005-03-17 08:19:48
|
When I query something like this: /databases/1/items?query='daap.songartist:Beatles','daap.album:da greeeat album'&output=xml I get a query with "or" between the artist and album. How do I make it "and" /a |
From: <and...@us...> - 2005-03-16 20:10:46
|
I've just briefly tried the sqlite driven mt-daapd and it works! :) Great! So this is what you can do on the "Playlist" tab in the web interface. Click on "Library" will load your whole library and show it as a table. If you click on the genre/artist/album boxes they will update and the song list will reflect the selection. Some questions: databases/1/browse should return the results alphabetically (I guess it is just a matter of adding “order by” to the select clauses.) (not a question I know) I can’t query some thing like this 'daap.songgenre:Children's Music' I encode the ' with %27 Itunes sorts artists that begin with "the" on the name e.g. “The Corrs” will be under C not T. Should we do the same? Can I through the daap queries tell the server how I want the result sorted? "Gimme all Jackson songs sorted by song length" /a |
From: Ron P. <ro...@pe...> - 2005-03-11 07:10:03
|
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 >> I hope to have everything compiling and working tomorrow. > > It all sounds wonderful. Can't wait to try it! In cvs and on the nightlies page (http://nightlies.mt-daapd.org) is the sqlite version. A crippled, barely working sqlite version, but... Some warnings: I don't have sync writes turned off, or indexes turned off, and there is still something like 4 queries per insert on mp3 scanning. Ugh. Sloooow. - - browse/query/index is broken -- won't work against a soundbridge - - compression is broken - - it probably leaks memory like a sieve - - xml output is broken basically everything is broken except iTunes, and that only works after it's taken a hideously long time to scan the mp3 files. :) Anyway, there is plenty to work on now! -- Ron -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.3 (Darwin) iD8DBQFCMURODl/VbZJe82cRAovAAJ0TN8cs3JiwMx3N6MjvgLAG9PlJjgCeMzq4 CJILWHVb4kBs9StZhegCNJ4= =UdcT -----END PGP SIGNATURE----- |
From: Ron P. <ro...@pe...> - 2005-03-11 03:28:45
|
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 > Why does the mp3:s encoded with itunes mp3 vbr show up as 64 kbit in > mt-daapd? > (I use debian testing with libid3tag0 0.15.1b-4) My stupidness. I thought that if a vbr file had an xing tag, then the encoder would put average bitrate in the first frame header. Obviously not. Average bitrate is easy enough to figure when you have an xing tag, since you know the length and number of samples in the output, and can calculate average bitrate exactly from the song length. I just didn't realize I had to. Yours is the second vote for a fix I've seen. :) -- Ron -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.3 (Darwin) iD8DBQFCMRBvDl/VbZJe82cRAgzYAJ9Lb9+/4JHCzFeRQZEBNXQ2RfQe4ACg4Lai jWnFPA1EOiHaPYTHBnIoBZA= =bw76 -----END PGP SIGNATURE----- |
From: <and...@us...> - 2005-03-10 19:10:54
|
Why does the mp3:s encoded with itunes mp3 vbr show up as 64 kbit in mt-daapd? (I use debian testing with libid3tag0 0.15.1b-4) /a |
From: Hamish A. <ha...@gm...> - 2005-03-10 17:50:46
|
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 > I hope to have everything compiling and working tomorrow. It all sounds wonderful. Can't wait to try it! Best wishes, Hamish -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.3 (Darwin) iD8DBQFCMIjg5+5l0g2lF18RAiXfAJ9DtAH6/ckThjOQegGAIyoZEK7gKACfbnr6 aP3M9SyS2E93ZZvrI1HTe8g= =+kHn -----END PGP SIGNATURE----- |
From: Timo J. R. <tr...@ik...> - 2005-03-10 08:14:33
|
Ron Pedde wrote: > The reason I haven't check anything in for a while is that I'm doing > that. I've ripped out playlists in favor of sql where clauses. The lex > and yacc stuff is gone. The playlist.[ch] is gone. The database now > generates dmap blocks directly rather than recordset->mp3file->dmap > tree. one malloc rather than ~40 strdups. daap.[ch] is gone. > daap-proto.[ch] is gone. db-memory is gone. db-gdbm is gone. the > red/black tree stuff is gone. Hi Ron Just what I would have done. Cool! -- Timo J. Rinne ** tr...@ik... ** http://www.iki.fi/tri/ |
From: Ron P. <ro...@pe...> - 2005-03-10 01:04:30
|
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 > I did some tests with sqlite. Didn't get mt-daapd actually work with > sql backend, but searches (plainly typed to sqlite command line ui) of > my 15000+ songs in different orders were very fast indeed. I think we > should nuke other db backends and optimize the database access logic > towards sql. The reason I haven't check anything in for a while is that I'm doing that. I've ripped out playlists in favor of sql where clauses. The lex and yacc stuff is gone. The playlist.[ch] is gone. The database now generates dmap blocks directly rather than recordset->mp3file->dmap tree. one malloc rather than ~40 strdups. daap.[ch] is gone. daap-proto.[ch] is gone. db-memory is gone. db-gdbm is gone. the red/black tree stuff is gone. Right now the database stuff works. Scanning works. I can't log in from an iTunes client yet, because I still have some problems in the dmap generation. I expect that once the database is built, then it will be pretty darn fast. And I think that the initial database building could be optimized for the "full rescan" case, and a little worse performance in a rescan wouldn't be noticed. I hope to have everything compiling and working tomorrow. -- Ron -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.3 (Darwin) iD8DBQFCL50gDl/VbZJe82cRAgOrAJ9Bc1YiY9y/zVV8P0ZOmyZ+yG10AACfTNhQ 94gxuSdFFB36kOiAiZOYiJM= =DDQq -----END PGP SIGNATURE----- |
From: Timo J. R. <tr...@ik...> - 2005-03-09 11:08:08
|
>>>> I think that even longer initial rescan is easy to bear if you can run >>>> it explicitly and follow the progress while running it. >>> Indeed. >> that makes a lot of sense, too. > Doesn't it? ;) http://www.iki.fi/tri/mt-daapd/patch/mt-daapd-cvs-20050305-initial-scan-only.patch Adds -S command line option that does initial scan in foreground in somewhat verbose way and exits. BTW, I'd still like song_sample_count or something like that to be added to the database. That would be very useful for server side conversion. I did some tests with sqlite. Didn't get mt-daapd actually work with sql backend, but searches (plainly typed to sqlite command line ui) of my 15000+ songs in different orders were very fast indeed. I think we should nuke other db backends and optimize the database access logic towards sql. -- Timo J. Rinne ** tr...@ik... ** http://www.iki.fi/tri/ |
From: <and...@us...> - 2005-03-08 21:07:55
|
> > I guess you could scan the string and verify that it was a valid utf-8 > string, but what do you do when it's not? Display a warning or > something? or replace the bad characters with Xs or something? > Wouldn't it be a good idea to bitch about it when the scan is performed or will it slow down the scan even more? Maybe just dump a list of the files with crappy tags. Or a nicer list in the web UI. > It's probably going to be a couple days before I can do this. I've > started the rip-and-replace on the database code, and it has > ramifications all over the place. Currently nothing is working, and > this with a pretty fair amount of work on it over the weekend. > > It looks like its going to be a step backward before it's a step forward. > As long as we're heading in the right direction (which I really think the database stuff is) I'm happy. I will dive into mid term coma for a while myself, so my updates will slow down a little for a while. /a |
From: Ron P. <ro...@pe...> - 2005-03-08 02:44:38
|
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 > We must come up with a way to prevent the mt-daapd server from sending > malformed utf-8. I've done some testing with wrongly tagged songs > where they get sent claiming to be utf-8 but aren't. The browser > refuses to build the xml document from the data. Right now you'll > notice that in the UI by the absence of songs, no error message, just > empty boxes and lists. So the server has to make sure it really sends > utf-8. I don't know, maybe just throw away the garbage characters. > Would that be a hard thing to implement? The thing I've found on these is that the tags claim to be latin1 or unicode but are really native codepage. Grip, for example, tags in native code page but sets the encoding type to latin1. That works fine on xmms, that assumes that latin1 means "native codepage", but as soon as you try and do something with it (like convert it to utf-8), then it blows up. I guess you could scan the string and verify that it was a valid utf-8 string, but what do you do when it's not? Display a warning or something? or replace the bad characters with Xs or something? > And, yeah, I'll make the song list scrollable and a thousand other > things... It's probably going to be a couple days before I can do this. I've started the rip-and-replace on the database code, and it has ramifications all over the place. Currently nothing is working, and this with a pretty fair amount of work on it over the weekend. It looks like its going to be a step backward before it's a step forward. -- Ron -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.3 (Darwin) iD8DBQFCLRGWDl/VbZJe82cRAmzaAJkBHr+IPUqNUq2AlAx1QEMi+qSL1ACff2b7 8a/Yg343hV89V+oxFeLufA8= =2XGf -----END PGP SIGNATURE----- |
From: Timo J. R. <tr...@ik...> - 2005-03-07 19:21:54
|
Hi I wonder how I got it to work before. Anyways it was totally broken. Small patch to fix it: http://www.iki.fi/tri/mt-daapd/patch/mt-daapd-cvs-20050305-match-to-work.patch -- Timo J. Rinne ** tr...@ik... ** http://www.iki.fi/tri/ |
From: <and...@us...> - 2005-03-07 10:46:05
|
We must come up with a way to prevent the mt-daapd server from sending malformed utf-8. I've done some testing with wrongly tagged songs where they get sent claiming to be utf-8 but aren't. The browser refuses to build the xml document from the data. Right now you'll notice that in the UI by the absence of songs, no error message, just empty boxes and lists. So the server has to make sure it really sends utf-8. I don't know, maybe just throw away the garbage characters. Would that be a hard thing to implement? And, yeah, I'll make the song list scrollable and a thousand other things... /a |
From: Timo J. R. <tr...@ik...> - 2005-03-06 00:51:19
|
Hi All Finally. It's here. David Hammerton made his free ALAC decoder available in the net earlier today. I integrated it to mt-daapd (current cvs-version) as server side conversion option. You can get alac-program from David's web site http://www.crazney.net/ Server side filter capable to serve ALAC and AAC files through mt-daapd's server side conversion functionality can be found in http://www.iki.fi/tri/mt-daapd/ This is very cool indeed! Regards, -- Timo J. Rinne ** tr...@ik... ** http://www.iki.fi/tri/ P.S. My web site moved just a second ago, and may not be available for an hour or so. Meanwhile you can use http://www.paavi.com/tri/mt-daapd/ as a shortcut. |
From: Timo J. R. <tr...@ik...> - 2005-03-06 00:46:35
|
Hi Just uptaded my FLAC libraries to version 1.1.0 and noticed that mt-daapd configure check no longer works. The new libFLAC seem to require also -lm in order to link nicely. This is somewhat inconvenient. Also I didn't immediately know how this check should be modified so that -lm would be added only if it is really needed. Maybe someone more familiar with autoconf magic could patch this. -- Timo J. Rinne ** tr...@ik... ** http://www.iki.fi/tri/ |
From: <and...@us...> - 2005-03-06 00:40:24
|
Hello list! I've just checked in the first version of the playlist editor. Everything is read only right now, you can't change any playlists. This is what should happen. (only tested with Firefox on windows) In the web interface there is a new tab "playlist". It will show your playlists under "Source" and it will populate the Genre, Artist and Album select boxes. If you click on a playlist you will get a listing of all the songs in the list. I was to lazy too divide the song time into minutes and seconds so that field is empty (you get it in millisecs from the server) I've just tried it out on a small song library. If you have a several thousands song library it would be great if you could test how it runs. When you select "Library" under "Source" you will get the whole library. Your browser might explode, be aware. All calls to the server are made syncronously, so the browser will hang until it has all the data it asked for. If it takes a minute to get and render the 11k songs library the browser will lock up for a minute. I will change the code to make async calls later. I guess this will roll into a nightly if you don't want to run the cvs version. /a |
From: Ron P. <ro...@pe...> - 2005-03-05 04:51:48
|
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 > By the way, has anybody tried using any database other than > "databases/1" yet? iTunes only looks at database 1. Even if you tell it you have multiple databases. I guess it's marked for future expansions or something. At least, that's how it worked last time I checked it. This was two or three iTunes versions ago. -- Ron -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.3 (Darwin) iD4DBQFCKTrdDl/VbZJe82cRAsTSAJdK8mZdVfIiPo8hh0V/GY2EhZx5AKDE0S/8 J3nePAduyfhUlLpjK/uMbQ== =mpGW -----END PGP SIGNATURE----- |
From: <and...@us...> - 2005-03-04 17:28:31
|
>I guess it's a per-user thing, which we could treat like a static >version of Ron's "upper(genre)not like %COUNTRY%" password-based >filtering. We'd have to find some way of communicating which tracks >were disabled with the server though -- probably easiest through the >web interface although, of course, we could use DAAP scripting to >achieve it through iTunes! > What happened to simple? I want my songs shared, I put them in the mt-daap directory. I don't want my songs shared, I put them somewhere else. About the scanning, it would be possible to show the scan progress in the html interface. Don't know if people prefere to stare at a command prompt or a web page. /a |
From: Hamish A. <ha...@gm...> - 2005-03-04 15:21:32
|
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 > In iTunes you can share either entire library and all playlists or just > a selected set of playlists. Each track in the library has a checkbox > that is basically an "enabled"-flag but it doesn't affect on sharing. > Thenagain in Roku for example you can choose to skip "unchecked tracks" > by configuration. In an iTunes share, if files are unchecked at the server end they are unchecked by default at the client end; the client can check and uncheck songs locally but these settings do not seem to persist through a disconnect and reconnect. I guess it's a per-user thing, which we could treat like a static version of Ron's "upper(genre)not like %COUNTRY%" password-based filtering. We'd have to find some way of communicating which tracks were disabled with the server though -- probably easiest through the web interface although, of course, we could use DAAP scripting to achieve it through iTunes! By the way, has anybody tried using any database other than "databases/1" yet? Best wishes, Hamish -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.3 (Darwin) iD8DBQFCKHyA5+5l0g2lF18RAhHPAJ48RiCTH2qdumJUZameUWJPRMkPbgCfVKTy 5QS7Daf7Qsp9AAyL2q8zsz0= =2FWY -----END PGP SIGNATURE----- |
From: Timo J. R. <tr...@ik...> - 2005-03-04 14:38:34
|
Patrick Dowling wrote: > I'd mostly agree if the db is being populated by scanning files, > although I can imagine having songs you never want to hear, but still > want in the library/directory for completeness' sake. Having them > somehow marked "do not share" instead of building playlists around > them would seem easier. > > Dunno if that's something anyone would actually use :) but I'm pretty > sure iTunes offers this type of feature. Maybe that's just something > to consider when filling the db from the iTunes xml. In iTunes you can share either entire library and all playlists or just a selected set of playlists. Each track in the library has a checkbox that is basically an "enabled"-flag but it doesn't affect on sharing. Thenagain in Roku for example you can choose to skip "unchecked tracks" by configuration. -- Timo J. Rinne ** tr...@ik... ** http://www.iki.fi/tri/ |
From: Patrick D. <pl...@gu...> - 2005-03-04 14:20:28
|
>> Except if they didn't want it in the database, why did they put in into >> the directory covered by the daap server? >> >> I'd argue that it should just be added. I'd mostly agree if the db is being populated by scanning files, although I can imagine having songs you never want to hear, but still want in the library/directory for completeness' sake. Having them somehow marked "do not share" instead of building playlists around them would seem easier. Dunno if that's something anyone would actually use :) but I'm pretty sure iTunes offers this type of feature. Maybe that's just something to consider when filling the db from the iTunes xml. The question then is if disabled songs should it just marked (i.e. have an "enabled" field) or just removed from/not added to the db? Cheers, -- Pat |
From: Timo J. R. <tr...@ik...> - 2005-03-04 14:18:58
|
Ron Pedde wrote: > Except if they didn't want it in the database, why did they put in into > the directory covered by the daap server? > I'd argue that it should just be added. I agree. > But I guess this comes back to the question of how often you do a deep > scan. I think it's us, the developers who scan more. Others, not so much. > Maybe the startup scan should only be done when you have an empty > database -- the first startup. After that, why not just skip the > initial scan and rely on the background scanning (or a manual rescan > request) to find changes. Opinions? I'd say you are right. Still I'd like to see the explicit rescan flag. >>> I think that even longer initial rescan is easy to bear if you can run >>> it explicitly and follow the progress while running it. >> Indeed. > that makes a lot of sense, too. Doesn't it? ;) -- Timo J. Rinne ** tr...@ik... ** http://www.iki.fi/tri/ |
From: Hamish A. <ha...@gm...> - 2005-03-04 14:00:07
|
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 > Still need to figure how we update > though. I made some suggestions earlier in the thread. Are they in keeping with the rest of DAAP? > And what about config info? > > I'd like to move as much of the configuration as possible to the > database, and use the same web interface to edit the config completely. > > I'm not sure what the interface for that should be. Maybe that's part > of the /server-info query? WIth some custom metas? That sounds good to me. Best wishes, Hamish -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.3 (Darwin) iD8DBQFCKGnP5+5l0g2lF18RApvCAKCT1wWDPmcpcg10cI12TSSaXwp1HACfYmvY UrakrsyiwTpyI0A30gEirh4= =ILQJ -----END PGP SIGNATURE----- |