libgettextpo does not obsolete as msgmerge does

[Dwayne Bailey]

libgettextpo will output all data for an obsolete message including #.
and #: entries while msgmerge will only output msg* and #[space]
entries.

Take the following:

# Some comment
#. Extracted comment
#: obsoleteme:10
msgctxt "noun"
msgid "One"
msgstr "Een"

msgmerge x.po /dev/null

And you get:

# Some comment
#~ msgctxt "noun"
#~ msgid "One"
#~ msgstr "Een"

Both #. and #: are dropped

If you take them same and set obsolete status using
po_message_set_obsolete then the following is created:

# Some comment
#. Extracted comment
#: obsoleteme:10
#~ msgid "One"
#~ msgstr "Een"

Both #. and #: remain.

I would expect the libgetextpo behaviour to follow that of msgmerge in
dropping the additional information.

The problem I'm facing is that I'd like my output to match that of
msgmerge but don't want to have to adjust the behaviour on an adhoc
basis to match msgmerge.  With any new feature the output would need to
be revisited to ensure that it matches gettext behaviour.

Two alternative suggestions:

1) If a message status is set to obsolete then only output data that
would match that of msgmerge - you will only lose data that is not of
much use anyway, it follows msgmerge current behaviour
2) Alter msgmerge to output all data as currently happens in
libgettextpo - you get extra data that could be useful at some stage, it
means a lot of new data and behaviour changes for existing users.

1) sounds like the correct minimally invasive approach to me.

-- 
Dwayne Bailey
Translate.org.za

+27-12-460-1095 (w)
+27-83-443-7114 (cell)