Hello Rafaella, see my comments inline.
Rafaella Braconi a écrit : > Dear All, > > currently not all languages released by Sun are available as full > installations sets. This is due to resource constraints, storage, backup > capacity and such stuff. Although we are going to ramp up our hardware > to enable more installation sets in the future, we cannot increase the > number of languages with full installation sets in the short term. > > What we would therefore like to propose here is a set of criteria which > can equally be met by ALL languages (both Sun and non-Sun languages) > which would keep the number of full installation sets more or less > stable in the short term and would allow reasonable growth. Languages > which meet the following conditions could be built with full > installation sets: > > 1- the GUI translation must be at least 80% completed > 2- the Online translation must be at least 80% completed > 3- 1 official build approval must have been submitted to the QATrack > tool during the last 3 releases The point 3 is a little bit too strong imho. I'd go for 2 releases. > > For all the languages which meet all the 3 above criteria Sun will > provide official full installation sets. > New languages which provide 1 and 2 for the first time and ask > explicitly for full installation sets will get full installation sets. > However, if after 3 releases no QA approval is recorded in the QAtrack > tool, Sun will provide language packs only, if the translation status > still remains at 80% translation. If the translation status is less than > 80% Sun will stop providing the builds for that language. I agree with the principle, however, I think that this asks the question of server space for the builds and projects not qualified according to the criteria above. The question, I think is, what should we do about them? Will Sun still provide their langpacks? > > For all the ones who are interested to know how we will count the 80% > rate, Jörg has kindly provided the following information: > > ====== > to determine the translation percentage in the "Translation Statistics > 2", we collect all translated strings for a given language, count the > words in the English source strings for these translated strings, and > then compare the total number of translated words to the total number > for en-US. For determining what a "word" is, we use a similar algorithm > as the Unix "wc" utility which defines white spaces as a separators > between words. > > The current "Translation Statistics 2" cannot be used directly to > generate the numbers for OOo, as we e.g. have projects as instset_native > and instset_ooo inside the database, but do not differentiate between an > internal statistics and one for OOo. Frank is going to enable an easy > way for generating the data for the OOo-Wiki statistics page within the > next weeks. > ====== > Forgive me if I'm not an expert in this matter, but I guess that sometimes there cannot be a perfect matching between the english reference list and the localized glossary. What do you do with Chinese, Indic languages, Vietnamese, or more simply, when languages translate in two or three words one single english word? Thank you for your proposal, Charles. > Please me know if you support the above proposal by the end of this week. > > Kind Regards, > Rafaella > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
