[Debconf-devel] Bug#560323: debconf: document the fact that the TITLE command is deprecated

Frans Pop elendil at planet.nl
Thu Dec 10 22:05:30 UTC 2009

On Thursday 10 December 2009, Joey Hess wrote:
> Frans Pop wrote:
> > > (unless the string comes in via passthrough).
> >
> > And that is exactly where most "obsolete command" warnings currently
> > come from :-/
> And most of _those_ in turn are probably due to debconf communicating
> the default titles it sets ("Configuring <package>"), not due to a
> package calling TITLE.


> Maybe passthrough could use SETITLE for that, rather than TITLE. It
> would have to send along info about a pretend template that had a
> description consiing of the title value. Not sure about namespace issues
> for that, etc.

There's a comment about that in the passthrough code, but just removing the 
warning solves the problem as well.
Using TITLE in the passthrough frontend does seem natural enough. As long 
as we're keeping TITLE I see no real reason to make things more complex.

> From my grepping, a lintian test for "db_title <hardcoded string here>"
> would be useful.

Agreed. I'll file a bug report.
Although there will be false positives (htdig for example has a string that 
needs no translation), and even a variable can contain an untranslated 
string of course.

> > But if you feel there are valid use cases for TITLE I'm not against
> > keeping it, but in that case I think the deprecation warning in
> > cdebconf should be removed. I think D-I developers are sufficiently
> > aware of translation issues that they'll not use it.
> I have no problem with removing that warning.


How about the attached modified patch then? Instead of warning that TITLE 
is deprecated, it now explains why SETTITLE is normally to be preferred.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: debconf.patch
Type: text/x-diff
Size: 1504 bytes
Desc: not available
URL: <http://lists.alioth.debian.org/pipermail/debconf-devel/attachments/20091210/0fd893c7/attachment.patch>

More information about the Debconf-devel mailing list