Pootle, Virtaal & The Translate Toolkit

Very interested.  As suggested on #pootle I'd love to see some kind of
plan to migrate.  My ideal is a scenario where we see the patches being
logical and obvious and allows us to easily review the migration.

A branch is logical but I would love to see a situation where we can
merge code as we go.  Ie using both frameworks (if possible, it might
not be) and moving logical bits across.

On Mon, 2006-09-25 at 19:50 +0200, Zejn Gasper wrote:
> Hello, fellow Pootle users!
> 
> I've started to port Pootle from jToolkit to Django[1].
> 
> I will briefly try to explain why this is necessary.
> 
> When I started to hack around Pootle, I was more or less on my own.
> Documentation is very scarce, as there's only one tutorial on jToolkit's web
> page [2]. Current developers don't have much time to offer to newcomers, so 
> the
> only way was to dig into source code, browsing back and forth, to get known
> with source.
> 
> On the other side, Django is very well documented and it's developer community
> is quite big, making it easy to find help when a newcomer is stuck with
> framework. It is also more likely for a developer to contribute, if the 
> project
> has good documentation and by using Django, we'd have half of Pootle 
> documented
> very well. Also, Pootle developers now also maintain jToolkit, where if Pootle
> moves to Django, framework is maintained by others, leaving current Pootle
> developers more time for Pootle.
> 
> I intend to make migration with a set of patches against svn trunk. Maybe it
> would even be smart to create a new branch.
> 
> I'm now about 25% done and I've significantly compacted code, making it more
> readable. Some of the localization strings have also been moved to templates,
> which also makes some code obsolete.
> 
> What is your opinion on this?
> 
> Regards,
> Gasper Zejn
> 
> [1] http://www.djangoproject.com/
> [2] http://jtoolkit.sourceforge.net/tutorial.html
> 
> -------------------------------------------------------------------------
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to share your
> opinions on IT & business topics through brief surveys -- and earn cash
> http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> Translate-pootle mailing list
> Translate-pootle@...
> https://lists.sourceforge.net/lists/listinfo/translate-pootle
-- 
Dwayne Bailey
Translate.org.za

+27-12-460-1095 (w)
+27-83-443-7114 (cell)

Zejn Gasper wrote:
> Hello, fellow Pootle users!
>
> I've started to port Pootle from jToolkit to Django[1].
>
> I will briefly try to explain why this is necessary.
>
> When I started to hack around Pootle, I was more or less on my own.
> Documentation is very scarce, as there's only one tutorial on jToolkit's web
> page [2]. Current developers don't have much time to offer to newcomers, so 
> the
> only way was to dig into source code, browsing back and forth, to get known
> with source.
>
> On the other side, Django is very well documented and it's developer community
> is quite big, making it easy to find help when a newcomer is stuck with
> framework. It is also more likely for a developer to contribute, if the 
> project
> has good documentation and by using Django, we'd have half of Pootle 
> documented
> very well. Also, Pootle developers now also maintain jToolkit, where if Pootle
> moves to Django, framework is maintained by others, leaving current Pootle
> developers more time for Pootle.
>
> I intend to make migration with a set of patches against svn trunk. Maybe it
> would even be smart to create a new branch.
>
> I'm now about 25% done and I've significantly compacted code, making it more
> readable. Some of the localization strings have also been moved to templates,
> which also makes some code obsolete.
>
> What is your opinion on this?
>   
Hi

Since I'm the one who originally wrote Pootle using jToolkit, I thought
I should comment...

Because I wrote most of jToolkit (for my other work), using it meant
that I could get Pootle out the door initially without having to learn
any new framework.
Since that time, I've changed my mind on many of the design decisions
that were made for jToolkit (originally it didn't support any templates,
etc, which made the code REALLY ugly.). There have been some updates to
support templates and make life slightly easier (localizing strings
directly within the templates was one of the glaring uglinesses
remaining :-)) but I'm now working on a new web framework that revisits
these things.
Of course, nowadays there are much nicer web frameworks around like
Django, and so it makes sense for Pootle to move away from jToolkit.
I'm keen to see how this turns out - a branch would be a nice way to test it

David

Hello.

I have created a new page on translate wiki [1] for this effort. Everyone is
invited to give it a look, and more importantly, please specify what has to
work. Pootle probably has features still undiscovered by me and I don't want
them to break.

If everyone agrees, I'd like to ask for a creation of a new branch to 
migrate to
Django and access to Subversion repository. My username on sourceforge is
"zejn".

Regards,
Gasper Zejn

[1] http://translate.sourceforge.net/wiki/django_migration

----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.

On Sa, 2006-09-30 at 20:11 +0200, Gasper Zejn wrote:
> Hello.
> 
> I have created a new page on translate wiki [1] for this effort. Everyone is
> invited to give it a look, and more importantly, please specify what has to
> work. Pootle probably has features still undiscovered by me and I don't want
> them to break.
> 
> If everyone agrees, I'd like to ask for a creation of a new branch to 
> migrate to
> Django and access to Subversion repository. My username on sourceforge is
> "zejn".
> 
> Regards,
> Gasper Zejn
> 
> [1] http://translate.sourceforge.net/wiki/django_migration

Sorry for the late reply.  If you feel up to the task to do this, I'm
happy to support the creation of a branch.  I still think this will be a
big task, and it might require a lot of syncing from HEAD, but since you
are volunteering, I assume you are ready for all of this :-)

There is currently some refactoring and simplification going on. This
might already make a few things easier. Perhaps trying your hand at some
of the refactoring will help to get to know the code better, and to make
the migration easier. Either way, I think it will be good if you can
stay up to date with the developments on HEAD. Is there perhaps some
refactoring of the code that we should be doing first before the
migration?

If I understand the current page correctly, it will not really be
possible to test Pootle during the migration, and I think this might
make a smooth transition a bit hard, but please correct me if I'm wrong.

Is it possible to make a slightly more detailed description of the tasks
envolved in the migration?  This might make it easier to track progress,
or for other to join in. Just an idea.

F

On Tuesday 03 October 2006 08:06, F Wolff wrote:
> On Sa, 2006-09-30 at 20:11 +0200, Gasper Zejn wrote:
> > Hello.
> >
> > I have created a new page on translate wiki [1] for this effort. Everyone
> > is invited to give it a look, and more importantly, please specify what
> > has to work. Pootle probably has features still undiscovered by me and I
> > don't want them to break.
> >
> > If everyone agrees, I'd like to ask for a creation of a new branch to
> > migrate to
> > Django and access to Subversion repository. My username on sourceforge is
> > "zejn".
> >
> > Regards,
> > Gasper Zejn
> >
> > [1] http://translate.sourceforge.net/wiki/django_migration
>
> Sorry for the late reply.  If you feel up to the task to do this, I'm
> happy to support the creation of a branch.  I still think this will be a
> big task, and it might require a lot of syncing from HEAD, but since you
> are volunteering, I assume you are ready for all of this :-)

Yes.

> There is currently some refactoring and simplification going on. This
> might already make a few things easier. Perhaps trying your hand at some
> of the refactoring will help to get to know the code better, and to make
> the migration easier. Either way, I think it will be good if you can
> stay up to date with the developments on HEAD. Is there perhaps some
> refactoring of the code that we should be doing first before the
> migration?

That is a good idea. I've actually hit the point in porting, where it isn't a 
matter of toolkit but rather the current organisation of page classes.

> If I understand the current page correctly, it will not really be
> possible to test Pootle during the migration, and I think this might
> make a smooth transition a bit hard, but please correct me if I'm wrong.

While it might not be possible to test it directly, there is a tool I've 
started using, Selenium [1], that allows to check a webpage from within 
browser. I'll upload this somewhere now so you can test it.

> Is it possible to make a slightly more detailed description of the tasks
> envolved in the migration?  This might make it easier to track progress,
> or for other to join in. Just an idea.
>
> F

regards,
Gasper Zejn

[1] http://www.openqa.org/selenium/

On Sat, 2006-09-30 at 20:11 +0200, Gasper Zejn wrote:
<snip>
> If everyone agrees, I'd like to ask for a creation of a new branch to 
> migrate to
> Django and access to Subversion repository. My username on sourceforge is
> "zejn".

Added.  Feel free to add the branch.

-- 
Dwayne Bailey
Translate.org.za

+27-12-460-1095 (w)
+27-83-443-7114 (cell)

On Tue, 2006-10-03 at 09:58 +0200, Zejn Gasper wrote:
<snip>
> While it might not be possible to test it directly, there is a tool I've 
> started using, Selenium [1], that allows to check a webpage from within 
> browser. I'll upload this somewhere now so you can test it.

<snip>

This is very interesting as we lost some functionality in the migration
from old jToolkit pages to KID templates.  Functionality that I've only
noticed is missing as I try to admin some aspects of Transalte.org.za's
Pootle.

I wonder if it would be worth creating some sort of regression testing.
So that we can at least check that we have basic functionality present
after the migration?

> [1] http://www.openqa.org/selenium/

-- 
Dwayne Bailey
Translate.org.za

+27-12-460-1095 (w)
+27-83-443-7114 (cell)

<snip>
> I wonder if it would be worth creating some sort of regression testing.
> So that we can at least check that we have basic functionality present
> after the migration?
>

I've already made some tests with Selenium, for logging in and out and 
similar. At the moment I don't have much time (moving another project), but 
with some luck, it might get uploaded next week or so.

Here are my answers your comments on wiki page (I'm responding here on mail, 
to make sure everyone sees this.)

Yes, Django has localizable strings in templates, which get pulled from 
compiled po. You can also skim the documentation [1]. There are no ugly 
templates, no html to translate. In fact, there will hopefully be no 
additional strings or string changes.

Work around PootleUnit will probably not affect my work, I hope. I've also 
changed the plan a little. First, I'll refactor the code, to push some 
features into the backend or create a new file with useful functions. The 
main idea is to split the code in two pieces. Those that will stay around 
after migration and some wrappers to make it compatible till then. This is 
mostly true with 'getsomething' methods of PootlePage subclasses.

Why Django and why not TurboGears?
Partly the same reason as it was with jToolkit: I'm very familiar with it.  
Other reason I also like is keeping the number of dependencies low. With TG 
you need to install specific versions of subprojects, but with Django you 
just need Django, therefore you are a bit less tied. 

Regards,
Gasper Zejn

[1] http://www.djangoproject.com/documentation/i18n/

On Wed, 2006-10-04 at 16:17 +0200, Zejn Gasper wrote:
> <snip>
> > I wonder if it would be worth creating some sort of regression testing.
> > So that we can at least check that we have basic functionality present
> > after the migration?
> >
> 
> I've already made some tests with Selenium, for logging in and out and 
> similar. At the moment I don't have much time (moving another project), but 
> with some luck, it might get uploaded next week or so.
> 
> Here are my answers your comments on wiki page (I'm responding here on mail, 
> to make sure everyone sees this.)
> 
> Yes, Django has localizable strings in templates, which get pulled from 
> compiled po. You can also skim the documentation [1]. There are no ugly 
> templates, no html to translate. In fact, there will hopefully be no 
> additional strings or string changes.

Great!

> Work around PootleUnit will probably not affect my work, I hope. I've also 
> changed the plan a little. First, I'll refactor the code, to push some 
> features into the backend or create a new file with useful functions. The 
> main idea is to split the code in two pieces. Those that will stay around 
> after migration and some wrappers to make it compatible till then. This is 
> mostly true with 'getsomething' methods of PootlePage subclasses.

This sounds like a more reliable approach.  Once that is there then we
can slowly eliminate this layer.

> Why Django and why not TurboGears?
> Partly the same reason as it was with jToolkit: I'm very familiar with it.  
> Other reason I also like is keeping the number of dependencies low. With TG 
> you need to install specific versions of subprojects, but with Django you 
> just need Django, therefore you are a bit less tied. 

Thanks.  I just wondered how much though had gone into considering other
frameworks.  Since we are starting from a clean slate it makes little
difference but knowing it is well documented, adopted and maintained is
important.  I think Guido van Rossum gave it the thumbs up recently.  So
sounds good.



-- 
Dwayne Bailey
Translate.org.za

+27-12-460-1095 (w)
+27-83-443-7114 (cell)

Hi All,

I really like the idea of migrating to django - I have researched the
platform some time back (I actually hoped we might be able to use it
for the translate.org.za website at some stadium since it looks very
nice) but never got around to actually playing with it.

If we can throw kid out of the door that would also be really nice
IMHO because the installation of kid is somewhat complex since it has
dependencies and requires hacks [1] to get it to run sometimes.

Anyway, I am interested though to know how this will impact Pootle
development. Since I need to learn Python I am thinking about adding a
password recovery feature some time (if nobody beats me to it) but if
we have to start on a "clean slate", would putting serious effort into
building the current main branch be worth the time or should one
rather focus on getting things migrated to django? What would the
focus thus be at this point in time, or am I running too far ahead? :)

Regards,
Charl

[1] http://translate.sourceforge.net/wiki/pootle/installation#installation_on_microsoft_windows

-- 
Charl van Niekerk
http://charlvn.za.net

On Tuesday 10 October 2006 18:22, Charl van Niekerk wrote:
> Hi All,
>
> I really like the idea of migrating to django - I have researched the
> platform some time back (I actually hoped we might be able to use it
> for the translate.org.za website at some stadium since it looks very
> nice) but never got around to actually playing with it.

Django is quite fun. The project has loads of documentation and the framework 
has very good support for most of the features a developer needs. I actually 
can't remember one thing I would need and it does not feature. Some have been 
complaining about it's ORM model, but that's going to be sorted out with 
SQLObject branch soon. But since Pootle already has it's own storage backend, 
that does not present a problem.

> If we can throw kid out of the door that would also be really nice
> IMHO because the installation of kid is somewhat complex since it has
> dependencies and requires hacks [1] to get it to run sometimes.

Yes, Kid templating will be replaced by Django's templates in near future, 
hopefully. 

> Anyway, I am interested though to know how this will impact Pootle
> development. Since I need to learn Python I am thinking about adding a
> password recovery feature some time (if nobody beats me to it) but if
> we have to start on a "clean slate", would putting serious effort into
> building the current main branch be worth the time or should one
> rather focus on getting things migrated to django? What would the
> focus thus be at this point in time, or am I running too far ahead? :)
>
>
> Regards,
> Charl
>
> [1]
> http://translate.sourceforge.net/wiki/pootle/installation#installation_on_m
>icrosoft_windows

Frankly, I'm not sure. If you are already familiar with Pootle's code, then 
it's probably best not to wait. If you don't know Django yet, then maybe you 
could skim the tutorials first and then help migrating. I'm only afraid that 
during the migration, you might get confused if you don't know Django well 
yet. This is because I will try to make the migration as painless as 
possible, making sure not to leave repository in broken state. This will 
include mixing Django style with some Pootle's peculiarities. 

If it suits you, feel free to join. I'd be happy to help. :)

Regards, 
Gasper

On Di, 2006-10-10 at 18:22 +0200, Charl van Niekerk wrote:
> Hi All,
> 
> I really like the idea of migrating to django - I have researched the
> platform some time back (I actually hoped we might be able to use it
> for the translate.org.za website at some stadium since it looks very
> nice) but never got around to actually playing with it.
> 
> If we can throw kid out of the door that would also be really nice
> IMHO because the installation of kid is somewhat complex since it has
> dependencies and requires hacks [1] to get it to run sometimes.
> 
> Anyway, I am interested though to know how this will impact Pootle
> development. Since I need to learn Python I am thinking about adding a
> password recovery feature some time (if nobody beats me to it) but if
> we have to start on a "clean slate", would putting serious effort into
> building the current main branch be worth the time or should one
> rather focus on getting things migrated to django? What would the
> focus thus be at this point in time, or am I running too far ahead? :)
> 
> Regards,
> Charl
> 
> [1] http://translate.sourceforge.net/wiki/pootle/installation#installation_on_microsoft_windows
> 

I think we need to take into account how long the Django migration will
take. Perhaps Gasper can comment on a rough timeline that he expects.
My _feeling_ is that it will take quite some time, and therefore a new
feature or improvement on the current Pootle will probably ship with a
few versions before we do the migration. So I wouldn't say it is a waste
of time. So depending on when you have time available and how the Django
roadmap goes, it could be worthwhile to add to the current Pootle. From
now on we should design and implement features with the migration in
mind, to make it less work later.

F

On Wed, 2006-10-11 at 08:16 +0200, F Wolff wrote:
> On Di, 2006-10-10 at 18:22 +0200, Charl van Niekerk wrote:
> > Hi All,
> > 
> > I really like the idea of migrating to django - I have researched the
> > platform some time back (I actually hoped we might be able to use it
> > for the translate.org.za website at some stadium since it looks very
> > nice) but never got around to actually playing with it.
> > 
> > If we can throw kid out of the door that would also be really nice
> > IMHO because the installation of kid is somewhat complex since it has
> > dependencies and requires hacks [1] to get it to run sometimes.
> > 
> > Anyway, I am interested though to know how this will impact Pootle
> > development. Since I need to learn Python I am thinking about adding a
> > password recovery feature some time (if nobody beats me to it) but if
> > we have to start on a "clean slate", would putting serious effort into
> > building the current main branch be worth the time or should one
> > rather focus on getting things migrated to django? What would the
> > focus thus be at this point in time, or am I running too far ahead? :)
> > 
> > Regards,
> > Charl
> > 
> > [1] http://translate.sourceforge.net/wiki/pootle/installation#installation_on_microsoft_windows
> > 
> 
> I think we need to take into account how long the Django migration will
> take. Perhaps Gasper can comment on a rough timeline that he expects.
> My _feeling_ is that it will take quite some time, and therefore a new
> feature or improvement on the current Pootle will probably ship with a
> few versions before we do the migration. So I wouldn't say it is a waste
> of time. So depending on when you have time available and how the Django
> roadmap goes, it could be worthwhile to add to the current Pootle. From
> now on we should design and implement features with the migration in
> mind, to make it less work later.

With that in mind a feature with a Selenium test would be the best, as
then we can ensure that we don't regress feature wise.

-- 
Dwayne Bailey
Translate.org.za

+27-12-460-1095 (w)
+27-83-443-7114 (cell)

Zejn Gasper wrote:
> On Tuesday 10 October 2006 18:22, Charl van Niekerk wrote:
>   
>> If we can throw kid out of the door that would also be really nice
>> IMHO because the installation of kid is somewhat complex since it has
>> dependencies and requires hacks [1] to get it to run sometimes.
>>     
> Yes, Kid templating will be replaced by Django's templates in near future, 
> hopefully. 
>   
You mean these: http://www.djangoproject.com/documentation/templates/
They look really ugly to my eye :-)
I presume one can use Django without the templates...
I'm currently switched from using Kid to using Genshi which implements
almost the same syntax but much more cleanly:
http://genshi.edgewall.org/
Switching Pootle to Genshi would probably involve only a few lines of
code (change the import)
>> [1]
>> http://translate.sourceforge.net/wiki/pootle/installation#installation_on_m
>> icrosoft_windows
>>     
Hmmm .. I can provide an installer for Kid on Windows if anyone needs
this while kid is still the current choice:
http://davidf.sjsoft.com/files/kid-0.9.3.win32.exe

David

On Wednesday 11 October 2006 13:49, David Fraser wrote:
> Zejn Gasper wrote:
> > On Tuesday 10 October 2006 18:22, Charl van Niekerk wrote:
> >> If we can throw kid out of the door that would also be really nice
> >> IMHO because the installation of kid is somewhat complex since it has
> >> dependencies and requires hacks [1] to get it to run sometimes.
> >
> > Yes, Kid templating will be replaced by Django's templates in near
> > future, hopefully.
>
> You mean these: http://www.djangoproject.com/documentation/templates/
> They look really ugly to my eye :-)

Why do you think they're ugly? Is there something specific or just the general 
look of it? To me, kid looks ugly. :)

It has one feature that is both good and bad: it's not XML based. This means 
you must take more care to output valid XML and you are not tied to XML 
output, therefore it's easier to output text based formats, eg. CSV.
Oh, and it has inheritance unlike kid.

> I presume one can use Django without the templates...

Yes, Django tries not to tie you. You can use any other template language.

> I'm currently switched from using Kid to using Genshi which implements
> almost the same syntax but much more cleanly:
> http://genshi.edgewall.org/
> Switching Pootle to Genshi would probably involve only a few lines of
> code (change the import)

Hmm, nice. Is genshi already packaged for Debian?


Gasper

Zejn Gasper wrote:
> On Wednesday 11 October 2006 13:49, David Fraser wrote:
>   
>> Zejn Gasper wrote:
>>     
>>> On Tuesday 10 October 2006 18:22, Charl van Niekerk wrote:
>>>       
>>>> If we can throw kid out of the door that would also be really nice
>>>> IMHO because the installation of kid is somewhat complex since it has
>>>> dependencies and requires hacks [1] to get it to run sometimes.
>>>>         
>>> Yes, Kid templating will be replaced by Django's templates in near
>>> future, hopefully.
>>>       
>> You mean these: http://www.djangoproject.com/documentation/templates/
>> They look really ugly to my eye :-)
>>     
> Why do you think they're ugly? Is there something specific or just the general 
> look of it? To me, kid looks ugly. :)
>   
Of course, ugly is all a matter of taste here :-)
To me what stands out about Kid as being a helpful syntax is that it
strikes a balance between being quite HTML-ish (so that you feel like
you are actually writing a web page) and having some Pythonic idiom.
> It has one feature that is both good and bad: it's not XML based. This means 
> you must take more care to output valid XML and you are not tied to XML 
> output, therefore it's easier to output text based formats, eg. CSV.
> Oh, and it has inheritance unlike kid.
>   
Yes I'd like to look into the inheritance stuff more... I've done some
stuff to overlay inheritance on top of kid/Genshi.
Genshi also has a text-based variant - but it uses a slightly different
syntax to the XML one, which I think makes sense.
>> I presume one can use Django without the templates...
>>     
> Yes, Django tries not to tie you. You can use any other template language.
>   
Good
>> I'm currently switched from using Kid to using Genshi which implements
>> almost the same syntax but much more cleanly:
>> http://genshi.edgewall.org/
>> Switching Pootle to Genshi would probably involve only a few lines of
>> code (change the import)
>>     
> Hmm, nice. Is genshi already packaged for Debian?
>   
Not sure, but its pure python so it can't be hard :-)
I even have it running on Windows CE using PythonCE without any
modifications (fun, fun)

Cheers
David

On 10/11/06, David Fraser <davidf@...> wrote:
> Hmmm .. I can provide an installer for Kid on Windows if anyone needs
> this while kid is still the current choice:
> http://davidf.sjsoft.com/files/kid-0.9.3.win32.exe

Nice, can I update the wiki page to point to that?

I have been doing a little more research, and I have to be honest that
I am not entirely convinced anymore that Django is the way to go.

Firstly, in order to run Django it seems that you need a DBMS
(Postgres or MySQL). This is not unreasonable I think but the current
Pootle doesn't require something like that.

Secondly, other than the templating system, what will the real
advantages of Django be? In terms of templating, there are many
options, and outside of that, Django seems to be largely about
information handling. I assume all the translatable/translated strings
are now going to go into the Postgres / MySQL database?

This could be an interesting new move for Pootle, but it seems like
Django will change many of the core design decisions. Is everybody in
favour? Personally I am starting to think that Pootle should rather be
built around its own framework (which just needs to be cleaned out and
documented). Will it really be easier to put Pootle on Django rather
than to do the cleaning/documenting? Also, how will the
ease-of-installation be affected?

Just some food for thought. :)

-- 
Charl van Niekerk
http://charlvn.za.net

Quoting Charl van Niekerk <charlvn@...>:

> On 10/11/06, David Fraser <davidf@...> wrote:
>> Hmmm .. I can provide an installer for Kid on Windows if anyone needs
>> this while kid is still the current choice:
>> http://davidf.sjsoft.com/files/kid-0.9.3.win32.exe
>
> Nice, can I update the wiki page to point to that?
>
> I have been doing a little more research, and I have to be honest that
> I am not entirely convinced anymore that Django is the way to go.
>
> Firstly, in order to run Django it seems that you need a DBMS
> (Postgres or MySQL). This is not unreasonable I think but the current
> Pootle doesn't require something like that.

Django is a framework made for rapid web site developement. It features 
neat url
resolving, text based templating engine and "model" objects, that are 
mapped to
DB tables. And it has probably the most useful error pages.

Then it's up to developer to decide whether he will use django's templating
engine or genshi, django's DB mapping or SQLAlchemy.

> Secondly, other than the templating system, what will the real
> advantages of Django be? In terms of templating, there are many
> options, and outside of that, Django seems to be largely about
> information handling. I assume all the translatable/translated strings
> are now going to go into the Postgres / MySQL database?

Current Pootle code is very hard to read. Pages are classes, and there 
are even
pages, that have same methods, but those methods are intended to do completely
different things. With migration pootle would gain on code readability, since
pages are functions in Django and are far easier to track than hunting class
methods all over the code. Also, it would be nice to split code based on what
it does, eg. storage-api access in one file, common pootle settings in other,
etc.

As far as storage is concerned, that is to be handled by Gintautas' SOC 
project
and not be held in DB.

If I make a step further, I'd make the storage-api branch a new daemon and
access all translations through it. It would make translation locking far
easier, since just the storage-api needs to know what's locked, and others get
"editing denied, locked by X Y" errors.

> This could be an interesting new move for Pootle, but it seems like
> Django will change many of the core design decisions. Is everybody in
> favour?

Django is not going to change any of the design decisions. If it would, 
then the
migration would already be done.

> Personally I am starting to think that Pootle should rather be
> built around its own framework (which just needs to be cleaned out and
> documented). Will it really be easier to put Pootle on Django rather
> than to do the cleaning/documenting?

I disagree on this. Django is here, now, and all we need comes with it. We
really don't want to duplicate work.

> Just some food for thought. :)
>
> --
> Charl van Niekerk
> http://charlvn.za.net
>

I may be a bit more words than work, let's hope that changes.

regards,
Gasper Zejn