Package: po4a Version: 0.46-1 Severity: normal

[ Fixing this would cause a disruptive change, feel free to tag it as wontfix if you don’t intend to make disruptive changes. This is a long standing issue already noticed in perkamon that was not raised before because it would be disruptive, but #786525 made me think that if we are going to make disruptive changes anyway, maybe this issue is worth fixing too. ]

Hi,

Double spaces after a dot or a closing parenthesis are kept verbatim, and a newline after a dot or a closing parenthesis is also “translated” by a double space.

This causes, for example (from hwclock(8)):

[…] (for compatibility with .BR %clock (8)) as a decimal integer.

to be translated in the following PO stanza:

#. type: Plain text #: C/man8/hwclock.8:735 msgid "" "[…] (for compatibility with B<\%clock>(8)) as a decimal integer." ^^

On the other hand, man(1) viewers do *not* display two spaces after this closing parenthesis while reading “man -LC 8 hwclock”, so this behaviour is a mistake introduced by po4a.

Ditto, this string in the manpage:

recent calibration. Zero if there has been no calibration yet or […]

is translated in the following PO stanza:

"recent calibration. Zero if there has been no calibration yet or […]"

Even if the double space after a dot is kept by (broken) design*, the double space after the closing parenthesis is definitely wrong.

Again, fixing this issue will cause many string to be fuzzied and might not be desirable because of that, but should be widely announced if fixed.

Regards

David

* “(broken) design”: The GNU project advise to use double spaces after a final period (and they are indeed respected by man(1) interfaces). In English, as in French, the space following a final period should be a “long space”. Of course, since the man(1) interfaces usually display text using monospace fonts, it makes not much sense, and instead it became usual to use two spaces instead in some fields. One may argue that forcing doc writers to handle such hack would be stupid and that only the viewers should be patched to offer readers the text displayed as it should be, but unfortunately, it seems the other way around has been chosen. Anyway, it makes little or no sense for po4a to respect the number of spaces in the msgid, and probably not so much when creating manpages from msgstr, but I’m drifting into some considerations that may not reach consensus and thus I’ll stop.