Merge lp://qastaging/~robert-ancell/intltool/xml_numeric_charrefs into lp://qastaging/intltool
Proposed by
Robert Ancell
Status: | Rejected |
---|---|
Rejected by: | Данило Шеган |
Proposed branch: | lp://qastaging/~robert-ancell/intltool/xml_numeric_charrefs |
Merge into: | lp://qastaging/intltool |
Diff against target: |
25 lines (+4/-0) 2 files modified
intltool-extract.in (+2/-0) intltool-merge.in (+2/-0) |
To merge this branch: | bzr merge lp://qastaging/~robert-ancell/intltool/xml_numeric_charrefs |
Related bugs: |
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Данило Шеган | code | Needs Fixing | |
Review via email:
|
To post a comment you must log in.
Unmerged revisions
- 703. By Robert Ancell
-
Fix decimal charref regexp
- 702. By Robert Ancell <bob@alchemy>
-
make intltool recognise numeric character references (they occur in glade files)
Thanks for the fix, Robert. What is the reason to support hexadecimal digits in regular decimal entities (i.e. to have a-fA-F in a line for chr() call)?
Also, I am not sure we want this as such. What's the reasoning to convert them to real characters? Was it not possible to input them in the source document as real characters as well? intltool tries to do minimal conversion in these cases, allowing developers and translators full flexibility. I.e. you do convert these to actual characters, but never convert them back (though, that would be pretty bad because we can't tell which should be converted and which should not).
One big reason against it is that it might break many existing translations in the sense that it's an incompatible change (i.e. if someone has a message with these in, it would have to be re-translated to all the languages). Can you at least give the basic idea of how many of eg. GNOME projects are already using these entities and would be affected?
Also, what happens if intltool is not running in a Unicode locale and unsupported character for current locale is tried?
So far, I am leaning on rejecting this patch, but would like to hear more reasons behind it. For once, I know it'd be friendlier to translators in a general case, which is good except when you make them re-translate many things because of a toolchain change.
If we agree we want it, can you also add at least a minimal test case to tests/? I know it'd be best if we had unit tests for a thing like this, but since we don't, another test case added would be nice.