Menu
▾
▴
bodington-developers
|
From: Peter C. <Pet...@me...> - 2005-07-15 10:49:32
|
> From: bod...@li...=20 > I think the fact that templates for different languages get=20 > compiled is=20 > inherently duff. Conceptually, there should be one single=20 > generic set of=20 > templates that dynamically pull in the resource bundles at run-time. I thought that when I first saw the design, but have since moved to the opposite point of view. One of the reasons for compiling templates in the first place is efficiency, and we have a *lot* of templaces given the number of frames and framesets in Bod. I've come round to the view that it probably makes sense* to remove the calls to the property retrieval functions from the execution path of each page display. - Peter * This year. Moore's Law continues its apparently inexorable progress... |
|
From: Peter C. <Pet...@me...> - 2005-07-15 12:56:32
|
> From: bod...@li...=20 > it's only when you start doing=20 > something in anger that all the issues crystallise in one's=20 > (my) mind ;-). There's considerable evidence that software design is a wicked problem (http://en.wikipedia.org/wiki/Wicked_problems) - we have certainly seen parts 1, 2 and 3 of Conklin's four defining characteristics in the case of Bodington i18n; I'd like to think 4 won't apply in this case, but that has yet to be seen. - Peter |
|
From: Peter C. <Pet...@me...> - 2005-07-15 13:23:25
|
> From: [...] Alistair Young > If you want people to see what it's like give them PDFs! or install a > public version that they can login to and play with. This is one of the key marketing messages. Bodington is *not* easy to install and configure at the best of times. An out-of-the-box installation would help a lot, but brochureware may well help more. > It seems your concerns are addressed by the knoppix CD anyway. Dumb question: How often does the Knoppix CD *actually* work first time on a prospect's computer, without someone needing deep knowledge to install (say) display drivers? Is it easier to use than a vanilla Bodington war file, or harder? Especially if it prevents a prospect using their normal machine for its normal purpose while they're testing? - Peter |
|
From: Matthew B. <mat...@co...> - 2005-07-15 13:41:27
|
Peter Crowther wrote: >>It seems your concerns are addressed by the knoppix CD anyway. > > > Dumb question: How often does the Knoppix CD *actually* work first time > on a prospect's computer, without someone needing deep knowledge to > install (say) display drivers? Is it easier to use than a vanilla > Bodington war file, or harder? Especially if it prevents a prospect > using their normal machine for its normal purpose while they're testing? Knoppix has quite a large amount of clever autodetection (similar to a normal OS installation autodetection) that runs quickly at startup and selects the appropriate drivers. Problems with the Knoppix that I see the quickstart helps solve are: - Download size. Quickstart is 10MB. Knoppix is 700MB. - Dedicated PC. Knoppix requires that the PC stops being used for it's normal task. - Configuration. If Knoppix doesn't get the network detection correct then making the test install of Bodington accessible over the network could be a stumbling block. - Persistence. Knoppix doesn't have any persistence or easy way to upgrade to a full Bodington install. Also if the power goes so does your test data. - PC Only. Doesn't give Mac/Unix people a way to test. - Old. The Knoppix CD is harder to keep current with the latest Bodington. The big plus for the Knoppix is that if you have a PC you can spare for a little while it is almost guaranteed to work. The quickstart requires a little bit more work (but still allot less that a full install). I personally like the quickstart for the point of view a developer doing testing. I as a developer can get a clean Bodington install in 2 minutes without having to clear out databases or create another one or run setupservlet. -- +--Matthew Buckett-----------------------------------------+ | VLE Developer, Learning Technologies Group | | Tel: +44 (0) 1865 283660 http://www.oucs.ox.ac.uk/ | +------------Computing Services, University of Oxford------+ |
|
From: Peter C. <Pet...@me...> - 2005-07-15 13:42:57
|
> From: [...] Matthew Buckett > Has anyone got a few hours they can donate to me,=20 > I'm running a little short ;-) Donate, no - worse luck. Sell, maybe :-). - Peter |
|
From: Matthew B. <mat...@co...> - 2005-07-15 11:08:23
|
Peter Crowther wrote: >>From: bod...@li... >>I think the fact that templates for different languages get >>compiled is >>inherently duff. Conceptually, there should be one single >>generic set of >>templates that dynamically pull in the resource bundles at run-time. > > > I thought that when I first saw the design, but have since moved to the > opposite point of view. One of the reasons for compiling templates in > the first place is efficiency, and we have a *lot* of templaces given > the number of frames and framesets in Bod. I've come round to the view > that it probably makes sense* to remove the calls to the property > retrieval functions from the execution path of each page display. So are we going to distribute Bodington precompiled for all the available languages? If so the size of the distribution will become huge (compiled versions of english are 5.5MB on my machine). If not then we are changing Bodington so that it is no longer possible to deploy on a machine without a JDK. -- +--Matthew Buckett-----------------------------------------+ | VLE Developer, Learning Technologies Group | | Tel: +44 (0) 1865 283660 http://www.oucs.ox.ac.uk/ | +------------Computing Services, University of Oxford------+ |
|
From: Alistair Y. <ali...@sm...> - 2005-07-15 11:18:29
|
> If not then we > are changing Bodington so that it is no longer possible to deploy on a > machine without a JDK. yep. that's bod though. What other web application *compiles* it's own pages? Don't you need a JDK anyway? What happens when XmlTemplate decides the class file is out of date and tries to recompile the template? By allowing the absence of the JDK you're telling users "you may never customise your templates". Alistair --=20 Alistair Young Senior Software Engineer UHI@Sabhal M=F2r Ostaig Isle of Skye Scotland > Peter Crowther wrote: >>>From: bod...@li... >>>I think the fact that templates for different languages get >>>compiled is >>>inherently duff. Conceptually, there should be one single >>>generic set of >>>templates that dynamically pull in the resource bundles at run-time. >> >> >> I thought that when I first saw the design, but have since moved to th= e >> opposite point of view. One of the reasons for compiling templates in >> the first place is efficiency, and we have a *lot* of templaces given >> the number of frames and framesets in Bod. I've come round to the vie= w >> that it probably makes sense* to remove the calls to the property >> retrieval functions from the execution path of each page display. > > So are we going to distribute Bodington precompiled for all the > available languages? If so the size of the distribution will become hug= e > (compiled versions of english are 5.5MB on my machine). If not then we > are changing Bodington so that it is no longer possible to deploy on a > machine without a JDK. > > -- > +--Matthew Buckett-----------------------------------------+ > | VLE Developer, Learning Technologies Group | > | Tel: +44 (0) 1865 283660 http://www.oucs.ox.ac.uk/ | > +------------Computing Services, University of Oxford------+ > > > ------------------------------------------------------- > SF.Net email is sponsored by: Discover Easy Linux Migration Strategies > from IBM. Find simple to follow Roadmaps, straightforward articles, > informative Webcasts and more! Get everything you need to get up to > speed, fast. http://ads.osdn.com/?ad_id=3D7477&alloc_id=3D16492&op=3Dcl= ick > _______________________________________________ > Bodington-developers mailing list > Bod...@li... > https://lists.sourceforge.net/lists/listinfo/bodington-developers > |
|
From: Alexis O'C. <ale...@co...> - 2005-07-15 11:26:24
|
Alistair Young wrote: >>If not then we >>are changing Bodington so that it is no longer possible to deploy on a >>machine without a JDK. > > yep. that's bod though. What other web application *compiles* it's own > pages? Don't you need a JDK anyway? What happens when XmlTemplate decides > the class file is out of date and tries to recompile the template? > > By allowing the absence of the JDK you're telling users "you may never > customise your templates". > > Alistair > > Funnily enough I agree with Alistair on this one ;-). Requiring a JDK is reasonable enough for these very reasons. It would be cool if we could get the Eclipse JDT compiler of Tomcat 5.5 to compile Bodington templates. It was the way my thinking was going with the CompilerFacade to cater for this (and jikes) but I backed out (until such time) such wrappers for these compilers were fully implemented. (In the short-term, just wrapping javac with a facade seemed like obfuscatory overkill!). Alexis |
|
From: Alistair Y. <ali...@sm...> - 2005-07-15 11:09:04
|
> I thought that when I first saw the design, but have since moved to the > opposite point of view. you got the cheque then ;) We had a debate ages ago where either the language specific template classes had the strings from the Localiser hard coded in them, or they ha= d calls to the Localiser instead. It was decided that the dynamic method would be best as you could then update the translations without having to recompile the templates. It als= o meant I didn't have to do more code to tie properties files into the timestamp checking process of the template compiler. I really can't see what the problem is. XmlTemplate is asked to return a template instance, which is what it's always done. If the template class doesn't exist or it's out of date, XmlTemplate compiles it. Again, it's always done that. All that happens now is it adds a language variable int= o the process. So if you're a user with their language preference as "English" you get template_thingy_en.class. If you're a user with "Russian" you get template_thingy_ru.class What Alexis would prefer is English and Russian users get template_thingy.class which contains logic to query the User object, get their language preference and load the appropriate strings. I can't see how that improves on the current design. Every page access would then have the extra overhead of all that User centric logic. Alistair --=20 Alistair Young Senior Software Engineer UHI@Sabhal M=F2r Ostaig Isle of Skye Scotland >> From: bod...@li... >> I think the fact that templates for different languages get >> compiled is >> inherently duff. Conceptually, there should be one single >> generic set of >> templates that dynamically pull in the resource bundles at run-time. > > I thought that when I first saw the design, but have since moved to the > opposite point of view. One of the reasons for compiling templates in > the first place is efficiency, and we have a *lot* of templaces given > the number of frames and framesets in Bod. I've come round to the view > that it probably makes sense* to remove the calls to the property > retrieval functions from the execution path of each page display. > > - Peter > > * This year. Moore's Law continues its apparently inexorable > progress... > > > ------------------------------------------------------- > SF.Net email is sponsored by: Discover Easy Linux Migration Strategies > from IBM. Find simple to follow Roadmaps, straightforward articles, > informative Webcasts and more! Get everything you need to get up to > speed, fast. http://ads.osdn.com/?ad_idt77&alloc_id=16492&op=3Dclick > _______________________________________________ > Bodington-developers mailing list > Bod...@li... > https://lists.sourceforge.net/lists/listinfo/bodington-developers > |
|
From: Alexis O'C. <ale...@co...> - 2005-07-15 11:20:13
|
> > What Alexis would prefer is English and Russian users get > template_thingy.class which contains logic to query the User object, get > their language preference and load the appropriate strings. > > I can't see how that improves on the current design. Every page access > would then have the extra overhead of all that User centric logic. > > Alistair > Exactly! I think the performance issue is a misnomer. Sure we'd have to profile it, but in the general scheme of things we find that at Oxford Bodington spends relatively very little time in the servlet container and lots of time in the database. You're not still using Tomcat 4 instead of 5 are you? The performance increase is considerable ;-). Alexis |
|
From: Alistair Y. <ali...@sm...> - 2005-07-15 11:30:18
|
So what about, for 2.8: Template classes that query User.getLanguage() at run time. Localiser chain with the preferred language delegating to the default language for missing strings? This just isn't possible for 2.6. Maybe it might have been if we'd had this debate at the start of the process ;) > You're not still using Tomcat 4 no no! I'm just forever thinking of optimisations. A hangover from my real-time days. Alistair --=20 Alistair Young Senior Software Engineer UHI@Sabhal M=F2r Ostaig Isle of Skye Scotland >> >> What Alexis would prefer is English and Russian users get >> template_thingy.class which contains logic to query the User object, g= et >> their language preference and load the appropriate strings. >> >> I can't see how that improves on the current design. Every page access >> would then have the extra overhead of all that User centric logic. >> >> Alistair >> > > Exactly! > > I think the performance issue is a misnomer. Sure we'd have to profile > it, but in the general scheme of things we find that at Oxford Bodingto= n > spends relatively very little time in the servlet container and lots of > time in the database. > > You're not still using Tomcat 4 instead of 5 are you? The performance > increase is considerable ;-). > > Alexis > > > ------------------------------------------------------- > SF.Net email is sponsored by: Discover Easy Linux Migration Strategies > from IBM. Find simple to follow Roadmaps, straightforward articles, > informative Webcasts and more! Get everything you need to get up to > speed, fast. http://ads.osdn.com/?ad_id=3D7477&alloc_id=3D16492&op=3Dcl= ick > _______________________________________________ > Bodington-developers mailing list > Bod...@li... > https://lists.sourceforge.net/lists/listinfo/bodington-developers > |
|
From: Alexis O'C. <ale...@co...> - 2005-07-15 12:47:44
|
Alistair Young wrote: > So what about, for 2.8: > > Template classes that query User.getLanguage() at run time. > > Localiser chain with the preferred language delegating to the default > language for missing strings? Yes on both scores for me! > > This just isn't possible for 2.6. Maybe it might have been if we'd had > this debate at the start of the process ;) Yes, I understand. I kind of had reservations at the back of my mind when it was being discussed, but's it's only when you start doing something in anger that all the issues crystallise in one's (my) mind ;-). Alexis |
|
From: Matthew B. <mat...@co...> - 2005-07-15 12:47:56
|
Alistair Young wrote: > So what about, for 2.8: > > Template classes that query User.getLanguage() at run time. > > Localiser chain with the preferred language delegating to the default > language for missing strings? > > This just isn't possible for 2.6. Maybe it might have been if we'd had > this debate at the start of the process ;) Sounds very sensible. Has anyone got a few hours they can donate to me, I'm running a little short ;-) -- +--Matthew Buckett-----------------------------------------+ | VLE Developer, Learning Technologies Group | | Tel: +44 (0) 1865 283660 http://www.oucs.ox.ac.uk/ | +------------Computing Services, University of Oxford------+ |
|
From: Alistair Y. <ali...@sm...> - 2005-07-15 13:02:46
|
> Has anyone got a few hours they can donate to me, I don't mind doing this. I know a lot more about the bod soup now than when I started i18n, so I'm happy to put that new info to use. Is 2.8 ok for youz? If the templates become too slow we can always rollback to the 2.6 way of doing it. Alistair --=20 Alistair Young Senior Software Engineer UHI@Sabhal M=F2r Ostaig Isle of Skye Scotland > Alistair Young wrote: >> So what about, for 2.8: >> >> Template classes that query User.getLanguage() at run time. >> >> Localiser chain with the preferred language delegating to the default >> language for missing strings? >> >> This just isn't possible for 2.6. Maybe it might have been if we'd had >> this debate at the start of the process ;) > > Sounds very sensible. Has anyone got a few hours they can donate to me, > I'm running a little short ;-) > > -- > +--Matthew Buckett-----------------------------------------+ > | VLE Developer, Learning Technologies Group | > | Tel: +44 (0) 1865 283660 http://www.oucs.ox.ac.uk/ | > +------------Computing Services, University of Oxford------+ > > > ------------------------------------------------------- > SF.Net email is sponsored by: Discover Easy Linux Migration Strategies > from IBM. Find simple to follow Roadmaps, straightforward articles, > informative Webcasts and more! Get everything you need to get up to > speed, fast. http://ads.osdn.com/?ad_id=3D7477&alloc_id=3D16492&op=3Dcl= ick > _______________________________________________ > Bodington-developers mailing list > Bod...@li... > https://lists.sourceforge.net/lists/listinfo/bodington-developers > |
|
From: Matthew B. <mat...@co...> - 2005-07-15 13:23:40
|
Alistair Young wrote: >>Has anyone got a few hours they can donate to me, > > I don't mind doing this. I know a lot more about the bod soup now than > when I started i18n, so I'm happy to put that new info to use. Is 2.8 ok > for youz? Yeah. Your suggested plan seemed fine. I agree that we shouldn't change it for the 2.6 release. > If the templates become too slow we can always rollback to the 2.6 way of > doing it. Yep. -- +--Matthew Buckett-----------------------------------------+ | VLE Developer, Learning Technologies Group | | Tel: +44 (0) 1865 283660 http://www.oucs.ox.ac.uk/ | +------------Computing Services, University of Oxford------+ |
|
From: Matthew B. <mat...@co...> - 2005-07-15 13:03:17
|
Alexis O'Connor wrote: > Alistair Young wrote: > >>> If not then we >>> are changing Bodington so that it is no longer possible to deploy on a >>> machine without a JDK. >> >> >> yep. that's bod though. What other web application *compiles* it's own >> pages? Don't you need a JDK anyway? What happens when XmlTemplate decides >> the class file is out of date and tries to recompile the template? >> >> By allowing the absence of the JDK you're telling users "you may never >> customise your templates". Yep. But if we are trying to make it easy for people to try Bodington I think this is acceptable. > Funnily enough I agree with Alistair on this one ;-). Requiring a JDK is > reasonable enough for these very reasons. If we are going to say people need a JDK this needs to be well documentation and Bodington needs to have a very clear error message when you try to run it on a machine without one (not a stack trace). I still think we should run without a JDK if they want too, especially with the quickstart WAR. > It would be cool if we could get the Eclipse JDT compiler of Tomcat 5.5 > to compile Bodington templates. Indeed. -- +--Matthew Buckett-----------------------------------------+ | VLE Developer, Learning Technologies Group | | Tel: +44 (0) 1865 283660 http://www.oucs.ox.ac.uk/ | +------------Computing Services, University of Oxford------+ |
|
From: Alistair Y. <ali...@sm...> - 2005-07-15 13:15:10
|
Bodington isn't designed to be run in an environment without a JDK. If yo= u want to shoehorn it into such an environment for the sake of ease of deployment you're opening a huge can of worms. If you want people to see what it's like give them PDFs! or install a public version that they can login to and play with. You can't give them a shaky halfway house that works for a defined period of time, or are you planning on circumventing the timestamp checking for such a deployment? It seems your concerns are addressed by the knoppix CD anyway. Alistair --=20 Alistair Young Senior Software Engineer UHI@Sabhal M=F2r Ostaig Isle of Skye Scotland > Alexis O'Connor wrote: >> Alistair Young wrote: >> >>>> If not then we >>>> are changing Bodington so that it is no longer possible to deploy on= a >>>> machine without a JDK. >>> >>> >>> yep. that's bod though. What other web application *compiles* it's ow= n >>> pages? Don't you need a JDK anyway? What happens when XmlTemplate >>> decides >>> the class file is out of date and tries to recompile the template? >>> >>> By allowing the absence of the JDK you're telling users "you may neve= r >>> customise your templates". > > Yep. But if we are trying to make it easy for people to try Bodington I > think this is acceptable. > >> Funnily enough I agree with Alistair on this one ;-). Requiring a JDK = is >> reasonable enough for these very reasons. > > If we are going to say people need a JDK this needs to be well > documentation and Bodington needs to have a very clear error message > when you try to run it on a machine without one (not a stack trace). > > I still think we should run without a JDK if they want too, especially > with the quickstart WAR. > >> It would be cool if we could get the Eclipse JDT compiler of Tomcat 5.= 5 >> to compile Bodington templates. > > Indeed. > > -- > +--Matthew Buckett-----------------------------------------+ > | VLE Developer, Learning Technologies Group | > | Tel: +44 (0) 1865 283660 http://www.oucs.ox.ac.uk/ | > +------------Computing Services, University of Oxford------+ > > > ------------------------------------------------------- > SF.Net email is sponsored by: Discover Easy Linux Migration Strategies > from IBM. Find simple to follow Roadmaps, straightforward articles, > informative Webcasts and more! Get everything you need to get up to > speed, fast. http://ads.osdn.com/?ad_id=3D7477&alloc_id=3D16492&op=3Dcl= ick > _______________________________________________ > Bodington-developers mailing list > Bod...@li... > https://lists.sourceforge.net/lists/listinfo/bodington-developers > |
×
Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.
Thanks for helping keep SourceForge clean.
X