[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: Some problematic strings in Subversion

From: Jens Seidel <jensseidel_at_users.sf.net>
Date: Mon, 20 Oct 2008 22:22:26 +0200

On Mon, Oct 20, 2008 at 07:34:35PM +0100, Stefan Sperling wrote:
> On Mon, Oct 20, 2008 at 05:22:34PM +0200, Jens Seidel wrote:
> > msgid "No delimiter after 'node_kind' in tree conflict description"
> >
> > Should these error messages really get translated? Can they occur without
> > manual editing of a repository (dump file, db, ...)?
> No, when this error occurs, either there's a bug in Subversion,
> or the working copy has been corrupted, or both.

Yep, I expected this. How often did it already happen that such a
low-level error message was posted? It is rare, right and in bug reports
you also prefer English strings.

So let's omit marking similar messages for translation in the future?
Would this be OK?

> > Please try to collect such messages into something as
> > "No delimiter after '%s' in tree conflict description"
> > to reduce work for translators.
> OK. Can you make a patch?

I'm not sure and other people can do it probably much faster :-)
A possible problem could be allocating memory for temporary strings.

> > #: ../libsvn_wc/tree_conflicts.c:329
> > msgid "Invalid tree conflict data in entries file, but no idea what went wrong"
> >
> > Should "entries" be translated or is it a filename (.svn/entries)? Quotes
> > are missing in the latter case.
> r33785.
> "The entries file" is a general name for .svn/entries.

Ah, here are other ones which I (bogusly?) translated ("Eintragsdatei") :-)

Stefan, if you don't like this translation feel free to use quotes as
well. Your decision.

> > #: ../libsvn_wc/tree_conflicts.c:383 ../svn/tree-conflicts.c:280
> > msgid "Bad node_kind in tree conflict description"
> > s/node_kind/'node_kind'/ (same for "operation", "action", "reason")
> OK. Can you make a patch?

Yes, I will add the quotes and directly commit it as trivial fix.

> > #: ../svn/tree-conflicts.c:57
> > #, c-format
> > msgid ""
> > " The update attempted to delete '%s'\n"
> > " (possibly as part of a rename operation).\n"
> >
> > It's much better to avoid the indentation for translation and later
> > to perform s/^/ /, s/\n[^$]/\n /. So it is easiely possible to
> > change the indentaion later.
> I don't understand.

Content and formatting should be separated if possible. I agree that
combining it simplifies usually the code (e.g. here) but using some
helper functions like indent_text(_("The update attempted..."), 2)
it is possible to handle this.

> Does this mean we have to change something in the code,


Again it isn't strongly required but at least it should be considered
(especially for more complex formatting).

Think about the following:

  There are approximately 10 translations of Subversion available and
  at least 50 indented messages. This makes at least 500 strings in
  total. Do you really expect that all these are handled correctly?

> > A few times there occurs "victim" and I have trouble translating it. A victim
> > is the item which caused a tree conflict, right? So a translation similar to
> > "reason" would make sense? Do you have a small example (add it as translator
> > comment in the source)?
> Quoting libsvn_wc/tree_conflicts.h:
> * For example, a file that is deleted by an update but locally
> * modified by the user is a victim of a tree conflict.
> Basically, the victim is in danger. What that means exactly depends
> on the tree conflict at hand. For example, local changes made to
> the victim may disappear if a commit was made from its tree conflicted
> state. Or the victim may disappear from the versioned tree altogether.
> Or the victim is not in the tree but should be, because we have a text
> delta for it (in case of files), or file or subdirectory additions (in
> case of directories).
> I hope this explains it. It's hard to come up with a general definition.
> Since I speak German, let me make a suggestion which does not have
> the same negative connotations as the literal translation "Opfer":
> "(vom Baumkonflikt) betroffenes Objekt"
> Would that do?

Yes, I think so ("Opfer" would clearly be wrong). Even "Baumkonflikt"
sounds strange but it is the direct translation and think it is OK.

"konfliktverursachendes Objekt" is maybe even better (but it is more
like "culprit" which is the opposite of "victim").

> > --- subversion/svnadmin/main.c (Revision 33718)
> > +++ subversion/svnadmin/main.c (Arbeitskopie)
> > @@ -337,7 +337,7 @@
> >
> > {"dump", subcommand_dump, {0}, N_
> > ("usage: svnadmin dump REPOS_PATH [-r LOWER[:UPPER]] [--incremental]\n\n"
> > - "Dump the contents of filesystem to stdout in a 'dumpfile'\n"
> > + "Dump the contents of the repository to stdout in a 'dumpfile'\n"
> 'svnadmin dump' really *only* dumps the filesystem! No config files,
> no locks, no hook scripts, etc.

Ah, the repository consists also of this, right.
Nevertheless each single revision describes a filesystem, svnadmin dump
stores more than just a filesystem. But if you don't think filesystem is
wrong it's fine for me as well.

> I'd strongly suggest we really make it dump the whole repository before
> changing this message.

To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org
Received on 2008-10-20 22:23:24 CEST

This is an archived mail posted to the Subversion Dev mailing list.