Another Swedish translation update

Dirk Hohndel dirk at
Wed Oct 17 10:21:19 PDT 2012

Linus Torvalds <torvalds at> writes:

> This updates the Swedish translation for the new message strings, and
> the date ordering in particular. It also fixes an incorrect
> translation (it incorrectly had translated dive-time as if it was the
> starting time of the dive, not the duration of the dive).

Let me guess - you looked at the German translation when doing Swedish?
Because I made the same mistake :-)

> Side note to other people using MUA's that corrupt whitespace:
> The way to fix that tends to be to do what I've done here: attach just
> the *patch* as an attachment, and do the description and sign-off at
> the top of the email. Then, you can add commentary like this, which
> should not be part of the commit message, after the "---" marker in
> the email. You can have the "From:" and "Subject:" stuff at the top of
> the email if you want to, but if you skip them like I have done here,
> the date and from headerts of the email itself will be used.
> If you have a really broken email client that doesn't even fill in
> your real name correctly - or if you are forwarding a patch from
> somebody else (or if you have multiple separate email addresses and
> you want to be credited to another one than the one you are using),
> then you do need to make that explicit with the From: line at the top
> of the email body.
> That should make it trivial for Dirk to apply the patches and get the
> proper description, without him having to edit your email.

Who would have thought that Linus would have good recommendations for
how to send stuff to a maintainer using git... :-) :-) :-)

Yes, this is certainly the preferred way to do this (or do pull requests
- either is fine). Especially the "additional text below the --- is
extremely useful. I finally slipped yesterday and didn't amend a commit
from Henrik with the result that we now have a butt-ugly commit entry
for that one :-(

Sending patches the way Linus described prevents that from happening.

> BTW, Dirk - the "make install" target now seems to install these
> things fine on Linux, but looking at a strace of actually running
> subsurface with "LANG=sv_SE.UTF-8", for some reason subsurface only
> seems to use "$(pwd)/./share/locale/$LANG/", and never
> looks into /usr/share at all (but it *does* look into /usr/share for
> the libc translations).
> As a result, "LANG=sv_SE.UTF-8 subsurface" only works from within the
> subsurface source directory, not from anywhere else. I suspect there
> is some link-time flag for where to look for localization that is
> missing. This is on F17.

I'm pretty sure that's because we tell gettext explicitly to look in

	path = subsurface_gettext_domainpath();
	bindtextdomain("subsurface", path);

I had foolishly assumed that this only /adds/ the path to the search
path for the .mo files, but what you say seems to indicate that this
becomes the exclusive search path. And worse, reading the manpage seems
to indicate that you can't hand it a number of directories to search in
- this appears to become the one and only directory it searches in.



More information about the subsurface mailing list