Hi all,
Kees mentioned today that he saw another project that modified its makefile to only install translations that were at least XX%. The idea being to avoid installing really poor quality translations. Do you guys think that'd be worthwhile to do for Inkscape too? (And is anyone interested in hooking in the appropriate logic to the makefiles?)
Alternatively, I know many of our translations are unmaintained since Sodipodi days, and the percent translated for many are very low (under 10%). Is it at all worthwhile to have these files in our codebase? If they are of no true use, why don't we just eliminate them entirely? This would remove cruft, and improve build/install time.
Bryce
Hi,
Kees mentioned today that he saw another project that modified its makefile to only install translations that were at least XX%. The idea being to avoid installing really poor quality translations. Do you guys think that'd be worthwhile to do for Inkscape too?
IMHO translation ratio is not enough to judge the value of a translation. A very important question is: which strings are translated. If someone doesn't translate esoteric stuff like extensions, Inkboard, command line options etc, his translation can still be of value to us. To really judge things, we must actually take a look (switch the locale to the language in question, launch Inkscape and look around).
We could also go the GIMP way: separate the less important stuff into another PO file -- this way translation ratios could be enough for deciding about quality.
Alternatively, I know many of our translations are unmaintained since Sodipodi days, and the percent translated for many are very low (under 10%). Is it at all worthwhile to have these files in our codebase? If they are of no true use, why don't we just eliminate them entirely?
For certain people (me included) it's easier to edit/correct an existing text than to start from scratch. Moreover, it can be motivating for some potential translators to see a bad quality translation (again, I'm a good example for this -- YMMV).
IMHO we should keep the translations available, but only compile those translations to binary format (MO) that were touched lately by the translator and meet the above mentioned criteria (visual test or percentage).
Arpad Biro
__________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com
Arpad Biro wrote:
Kees mentioned today that he saw another project that modified its makefile to only install translations that were at least XX%. The idea being to avoid installing really poor quality translations. Do you guys think that'd be worthwhile to do for Inkscape too?
IMHO translation ratio is not enough to judge the value of a translation. A very important question is: which strings are translated. If someone doesn't translate esoteric stuff like extensions, Inkboard, command line options etc, his translation can still be of value to us. To really judge things, we must actually take a look (switch the locale to the language in question, launch Inkscape and look around).
This can be misleading. For example I translated the 'el' package from 2% to 15% but most changes are invisible (much of my work was on the clonetiler.cpp file, visible only to someone who will attempt to clone something). Also, feedback for bad translations can be obtained even from partial translations. And a partial translation may motivate some people to finish it!
However, a partial translation, especially on the main screen looks really ugly and unprofessional.
IMHO the best option is to mark some translations as 'full' (aka. 80%+) and all others as 'partial'. The user is informed and can choose to do what he likes. This would probably require some alterations in the startup code and would require more complex makefile (like processing the .po files and passing the values to the compiler as defines).
We could also go the GIMP way: separate the less important stuff into another PO file -- this way translation ratios could be enough for deciding about quality.
This would help, so everyone can see what the most important strings are.
Alternatively, I know many of our translations are unmaintained since Sodipodi days, and the percent translated for many are very low (under 10%). Is it at all worthwhile to have these files in our codebase? If they are of no true use, why don't we just eliminate them entirely?
For certain people (me included) it's easier to edit/correct an existing text than to start from scratch. Moreover, it can be motivating for some potential translators to see a bad quality translation (again, I'm a good example for this -- YMMV).
It is actually much easier to resume a partial translation than to start it from scratch. The 'el' translation was there from sodipoli 0.25 (in 2001) and the original translator didn't answer my e-mails. These 60 lines that were translated and their comments really provided some help into how to translate inkscape terminology in Greek. From this experience, I can say that even a 2% translation helps.
- Spyros Blanas
Le samedi 10 juin 2006 à 01:01 -0700, Arpad Biro a écrit :
Hi,
Kees mentioned today that he saw another project that modified its makefile to only install translations that were at least XX%. The idea being to avoid installing really poor quality translations. Do you guys think that'd be worthwhile to do for Inkscape too?
IMHO translation ratio is not enough to judge the value of a translation. A very important question is: which strings are translated. If someone doesn't translate esoteric stuff like extensions, Inkboard, command line options etc, his translation can still be of value to us. To really judge things, we must actually take a look (switch the locale to the language in question, launch Inkscape and look around).
We could also go the GIMP way: separate the less important stuff into another PO file -- this way translation ratios could be enough for deciding about quality.
Strongly agree. Splitting "core strings" from "extensions" and "rest of the strings" may help improving the quality of the global translation of Inkscape by stibilizing the core work, splitting the workload and help showing some priorities.
Alternatively, I know many of our translations are unmaintained since Sodipodi days, and the percent translated for many are very low (under 10%). Is it at all worthwhile to have these files in our codebase? If they are of no true use, why don't we just eliminate them entirely?
For certain people (me included) it's easier to edit/correct an existing text than to start from scratch. Moreover, it can be motivating for some potential translators to see a bad quality translation (again, I'm a good example for this -- YMMV).
Also agree there : we've just had 2 examples : Korean and Dutch translations have been quite improved (from a few % to a few dozens %).
regards,
matiphas
participants (4)
-
Arpad Biro -
Bryce Harrington -
Matiphas -
Spyros Blanas