Re: [bug-gettext] msgmerge --diff-to-previous (feature request)

[Ineiev]

On 10/30/2011 05:19 PM, Alexander Potashev wrote:
> I'm afraid, some programs that use their own .po
> parsers might fail.

I see; I didn't realize that some people never edit PO files
with general text editor.

> What I have checked with the above msgid_diff/msgid_plural_diff lines:
> 1. Lokalize (not sure about the version, must be around KDE SC 4.7.*)
> with its own parser.
> It thinks of concatenated msgid_diff+msgid_plural_diff as of plain "#|
> msgid" and "#| msg_plurals". In your case it thinks that the previous
> original text for the message was "(diff to previous msgid) (diff to
> previous msgid_plural)" for both singular (msgid) and plural
> (msgid_plural) forms.

Probably all programs would be happy if msgmerge used something like "# |"
for diffs, but such lines would accumulate on every subsequent merging.

> 2. My custom program that uses po_file_read() from libgettextpo-0.18.1.1.
> po_file_read() reports 3 errors.

Perhaps the parser in libgettextpo should be updated, too (my patch doesn't
do it).

> > # If the diffs go into
> > > "previous" ("#|"-d) messages, that may be unexpected for some
> > > translators and software.
> > I thought they just would avoid --diff-to-previous; would it be better
> > if the diffs went into messages prefixed with "#<some-other-character>"?
>
> «#* msgid "(diff to previous msgid)"» gets accepted silently by
> Lokalize and even by libgettextpo-0.18.1.1.

Do they add it to human comments?

> May be "#*" already means something?

I don't think so; gettext documentation only lists "#.", "#:", "#,", "#|".