[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: comments/string changes and issues with dpkg's messages



On Thu, 18 Aug 2005, Eddy Petrisor wrote:

> On 8/17/05, Adam Heath <doogie@debian.org> wrote:
> > On Wed, 3 Aug 2005, Eddy Petrisor wrote:
> >
> > > On 8/3/05, Denis Barbier <barbier@linuxfr.org> wrote:
> > > > On Tue, Jul 26, 2005 at 05:10:38PM +0300, Eddy Petrisor wrote:
> > > > [...]
> > > > > 3) in some cases the messages are so cryptical that rephrasing them would be a
> > > > > much better solution than adding automatic comments:
> > > > >
> > > > > message:
> > > > > "unable to open tmpfile for vsnprintf"
> > > > >
> > > > > code:
> > > > >   ...
> > > > >   if (!file) {
> > > > >   file= tmpfile(); if (!file) ohshite(_("unable to open tmpfile for
> > > > > vsnprintf"));
> > > > [...]
> > > >
> > > > Most of these ohshite() messages are indeed very cryptic, and could IMO
> > > > be removed from PO file because they are debug messages which are not
> > > > intended for end users.
> > >
> > > But what happens if a user has problems and encounters such a message?
> > > Suppose (for this particular case) that /tmp is for some reason on a RO media.
> > >
> > > A localized message might indicate to the user that he should do
> > > something about this issue, but an English one could be helpless.
> > >
> > > What do you think? Am I missing something?
> >
> > In that case, dpkg shouldn't be using oshite, but some other error message.
> >
> ohshite does not localize the message itself
>
> <quote>
>  ohshite(_("unable to open tmpfile for....
> </quote>
>
> What do you mean? Or am I missing your point?

If some string passed to ohshite needs to be translated, then ohshite should
not be called; instead, a proper error should be returned to the calling
code(either with the exception framework, or thru an error return).

ohshite should really be for internal errors only.  Things that "Should Not
Happen(tm)".



Reply to: