Sometimes I see a red "invalid-markup" message in pages that use MathML incorrectly. This string is hard-coded in C++ in two places.

Here is a patch to use nsContentUtils. This should also fix bug 346224 (even if this one is not really an issue now). I've also removed the "LOCALIZATION NOTE" in the layout/mathml/*.properties files, since apparently they are no longer accessible by translators according to bug 346224.

Comment on attachment 433814 [details] [diff] [review] Add localization for MathML r-. The screenshot is in bug 125541, and we shouldn't have XXX comments in l10n. I didn't investigate on whether the comment is right or nor. Is the content on http://www.mozilla.org/projects/mathml/fonts still up-to-date? It might make sense to not expose that to l10n, but to have that as a second parameter instead. If we end up changing the url, it's a single place thing only.

(In reply to comment #2) > (From update of attachment 433814 [details] [diff] [review]) > r-. The screenshot is in bug 125541, and we shouldn't have XXX comments in > l10n. I didn't investigate on whether the comment is right or nor. > > Is the content on http://www.mozilla.org/projects/mathml/fonts still > up-to-date? It might make sense to not expose that to l10n, but to have that as > a second parameter instead. If we end up changing the url, it's a single place > thing only. No, the dialog box is no longer used, see the comments after https://bugzilla.mozilla.org/show_bug.cgi?id=130023#c27 Maybe it's better to remove the messages of this dialog and rewrite them later if the warning is added back.

What are we localizing then? That other bug was just listing more bugs.

(In reply to comment #4) > What are we localizing then? That other bug was just listing more bugs. As indicated in the title, the only thing we need to localize is invalid_markup=invalid-markup I've also moved the message from layout/mathml/mathfont.properties to this properties file (bug 346224) but it is not really necessary now the dialog is removed.

I'd consider localizing that string to be a WONTFIX, "invalid-markup" isn't "localizable" as it's not carrying any really usable feedback to the user, nor any message with a meaning to translate for the localizer. I welcome other bugs to fix that. I don't mind removing dead code or false comments, which most of this patch does, though.

(In reply to comment #6) > I'd consider localizing that string to be a WONTFIX, "invalid-markup" isn't > "localizable" as it's not carrying any really usable feedback to the user, nor > any message with a meaning to translate for the localizer. I welcome other bugs > to fix that. > > I don't mind removing dead code or false comments, which most of this patch > does, though. This patch is also preparing the possibility to localize information from the MathML engine. See for instance bug 553917 (message errors sent to error console)

I'm really only talking about exposing strings to l10n. "invalid-markup" isn't anything that's helpful UE, but a bug to be fixed. No need to expose that to l10n, it's useless in any language, as far as I'm concerned. Even less so things that require XXX comments. Thus r- on this patch.

OK, so let's mark this bug as wontfix then.