I've done the count the results:
Rail Aliev's file: 4519
Doko's file: 68975
4519 seems unlikely. The file itself has no less than 72963 lines, so it should be 36.000 rather than 4.500.
But the other file seems to be quite correct! So I will trust doko's script for the creation of further GSI's from our .po files.
Regards,
Erdal
2006/3/11, Rail Aliev <[EMAIL PROTECTED]>:
On 01:37, Sat 11.03.06, Erdal Ronahi wrote:
>Hi Rail, hi Matthias,
>
>I am writing to both of you (and CCing Pavel), because you were helping us
>to build Kurdish GSI packs with different scripts from the Rosetta files.
>
>Today I've reviewed the results. I found that one of the sdf files is
>22.8MB (Matthias = doko), the other one only
>12.0 MB (rail). Now I wonder if they are both equally usable or if one of
>them is correct and the other broken.
>
>Matthias has already build a langpack for the upcoming Ubuntu
>6.04distribution which seems to be working - at least it doesn't break
>things
>itself.
>
>Is there a way to test whether the GSI files created with your scripts are
>indeed correct?
I think it is (must be) very simple. Just count en-US and ku messages and
compare with right one.
$ cat en-US.sdf |awk -F'\t' '{print $10}'|grep en-US|wc -l
It must be 68935 for OOB680_m5 for example. Then count ku messages in your
generated sdf:
$ cat your.sdf |awk -F'\t' '{print $10}'|grep ku|wc -l
If you have 100% translated or all of untranslated messages substituted
with en-US messages you'll have the same message count.
--
Best regards,
Rail Aliev
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
