[Fedora-trans-ar] .PO &.txt?.. what makes PO Very Special!!

[Maha Helwa]

On 9/3/05, Mohammad Ghoniem <Mohammad.Ghoniem at univ-ubs.fr> wrote: 
> 
> Salâms,
> 
> We need to decide how the data will be stored in the database. I can
> imagine two scenarios :
> 1- keep each po file as is i.e. in one big chunck corresponding to one
> record in the database.

 NOOOOOOOOO.. i don't prefer it in this way.. we once load the important 
data into db.. or each time we update the PO files from CVS.. update PO 
files with the approved string only!!
What left is db transcation.. each modification on a string in any PO file 
that will add a new record in db table

2- split po files and have one record per message.

 YEs that's what i did.. 

In all cases, we should make the choices explicit and draw a database
> specification.
> 
> If each po file is stored in one chunk, we need a parser like the one I
> proposed lately to load the data into objects, before we can push the
> data in some web form for the translators to deal with.

 Yes sure we need a parser to extract the data from PO file and put it into 
one of db table.. that's has nothing to do with either u save the whole file 
as chunk data; CBlob or text or whatever datatype or each string as a single 
record in the table.. 

 Once the
> translator is done, we need to rebuild the po file before storing it in
> the databse again. Some work is still needed here.

 NO... what's the use of the db then if we r going to update the files 
again.. the db is the intermediate layer here.. we need it to hold all the 
transaction the translators already did.. and keep them into db till the 
maintainer decide which is going to stay 'Approved' and which will be 
discarded and deleted from db.. 
So when we should compile the PO file again.. that's one of the system admin 
roles.. each meanwhile.. he compiles the file.. so what will happened?! we 
will go through the db table and pickup only the approved strings and update 
them at their correct position in the PO file..
 My point is that we dont' parse the file each time the translator modifying 
a string or reflect the translation change in PO file not before an approval 
from the maintainer side! in between will be db transaction..

 I am not sure why
> Maha you suggested to have a numerical identifier for each message. Is
> this really necessary ?

 You mean the primary key or the order of the string is also unique numeric 
value..
the order string we will use it when we compile the approved string and put 
them in the right place at the file.. what's why we need it to put it in the 
first place while we parse the file... 

In the web form, some data must be editable (msgstr) while some need not
> (msgid and file header).

 Yes?! didn't get ur point..

Salâm
> 
> Mohammad
> 
> --
> Fedora-trans-ar mailing list
> Fedora-trans-ar at redhat.com
> http://www.redhat.com/mailman/listinfo/fedora-trans-ar
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.fedoraproject.org/pipermail/trans-ar/attachments/20050903/e76a2b84/attachment.html