On Jul 9, 2009, at 1:24 AM, Philip Olson wrote:
How are CVS Revision numbers used today? The best feature of po is
that translation text is marked as 'fuzzy' when the main text is
changed. If we could do the same with our docbook translation files
whenever the en file is committed then we have the same benefits
imo. I'm wondering if CVS Revision numbers could be replaced by a
'fuzzy' markup tag...?
When a translator updates a translation (a file), they write which
CVS Revision in EN it's synced to. So, later when the EN version is
updated we then know that the translated is outdated. And all of our
translation tools rely on the incremental nature of CVS revision
numbers.
A couple of options....
Create an md5 hash on the version you are translating and use that to
"uniquely" identify it. If the EN version is modified by even a single
space, the hash will change and you'll know the files are out of sync.
What you won't know is just how far out of sync (but does that really
matter?)
If using XSLT at some point, use generate-id() function to generate a
unique id of EN and use that to uniquely identify the version. As with
md5, any minor changes will cause a different id to be generated.
Neither of these is quite as elegant as progressive version numbers. I
suspect md5 would be the best option.
Ted Stresen-Reuter
http://tedmasterweb.com
(I'm on vacation, but this thread is quite fascinating)
