Developers of Sympa

[Olivier Salaun - CRU <address@concealed>]

Hi,

Chia-liang Kao wrote:
> On Tue, May 27, 2003 at 05:21:36PM +0200, Olivier Salaun - CRU wrote:
>>You've already gone half the way, it is encouraging us to go further.
> 
> actually more than half the way now! http://lists.elixus.org/wws is running
> tt2 native version of sympa with half-done tw translations.

I know but I meant half the way because :
        * I noticed some template structure problems and accents problems
        * of template notions that have no equivalent in TT2

This is a major change in Sympa that could bring unstability ; we need to
be very carefull and double check things. We also need to provide backward
compatibility for people who customized templates or at least migration
scripts.

>> [...]
>>Your work drives us to (1) and (2). Do you know of any existing tools that 
>>would help translators
>>work on .tt2 and .po files ?
> 
> yes, the xgettext.pl distributed with locale::maketext::lexicon extracts 
> the strings
> to translate from tt2 files, and merges with exsiting po files.
> and all other gettext toolchain can apply to the po files.

Fine.

>>I have a question about .po files : the english text are the index ; how 
>>about duplicates (the
>>same english text that would be translated differently in different 
>>contexts) ?
> 
> AFAIK, the same key is merged as one entry in po files. however, i think 
> it's a common need
> for people to manually tweak the translations according to the context, 
> maybe autrijus
> would know if there is existing adopted solutions.

I'd be interested in Autrijus point of view about this...
I noticed a us.po that has no default translation, I guess it allows changing 
english translations without
changing the corresponding gettext key in all other .po files.

>>Other questions about TT2 : the format is very close to our native template 
>>format which is
>>great. Is there an equivalent for our [STOPPARSE] [STARTPARSE] ? Is there a 
>>way to get
>>the parser output stored in memory (an array), instead of having it printed 
>>to file descriptor ?
> 
> the parse_tpl function I implemented in src/tt2native.pl
> in my svn repository handles all input and output described
> in the original src/parse.pl. so yes, it handles arrayref output
> and arrayref as a template to process.

Fine.

> about start/stop parse, I have tried to implement it in tt2, but
> there's more problem than expected. more detail is already in my last
> mail on the list. in short, there should be escape from the upper
> level, ie, when using mhonarc to create the file to be processed by
> any template system. otherwise it's easy to exploit the stop/start parse 
> things. 

I noticed the security problem you reported and we'll try to get rid of that 
ASAP.
I'll check in TT2 documentation for non-parsing blocks...

> the current mhonarc resources and tt2 templates in my
> svn repository already handles the above problem, by changing the 
> tt2 starttag and endtag to something mhonarc would automatically
> escape without any kludge.

You mean replacing [% and %]. What did you replace them with ?

> the svn repository is at http://svn.clkao.org/sympa/trunk/,
> stuff people might be interested:
> src/tt2native.pl      # the tt2 implmenetation of parse.pl
> wwsympa/tt2           # the templates
> po/wwsympa            # the po files
> src/etc/mhonarc-ressources 

We will definitely work on the migration from tpl/nls to tt2/gettext next 
july (I doubt we
get enough time in the meanwhile). Could you send us a tar of your modified 
sympa distribution
or BETTER a diff of Makefiles/scripts + a detailed migration notice.

Anyhow we will contact you when we'll be working on this.

Thanks.

-- 
Olivier Salaun
Comite Reseau des Universites