Hi Stephen ,
Stephen Kitt <skitt@debian.org> (2014-08-19):
> On Mon, Jul 07, 2014 at 03:56:21AM +0200, Cyril Brulebois wrote:
> > Modestas Vainius <modax@debian.org> (2013-12-29):
> >
> > thanks for the patch but I'm not convinced, see below:
> >
> > > --- a/debian/iso-scan.postinst
> > > +++ b/debian/iso-scan.postinst
> > > @@ -162,7 +162,7 @@ scan_device_for_isos() {
> > > elif [ "$look_subdirs" = 1 ]; then
> > > opt="-type f"
> > > fi
> > > - isolist=$(find $dir $opt -name "*.iso" -o -name "*.ISO" 2>/dev/null)
> > > + isolist=$(find "$dir" $opt -name "*.iso" -o -name "*.ISO" 2>/dev/null)
> >
> > This part is certainly OK; at least I can't think of a reason why that
> > wouldn't be a good thing.
> >
> > > TOPLEVEL_DIRS_COUNT=$(($TOPLEVEL_DIRS_COUNT + 1))
> > >
> > > for iso in $isolist; do
> >
> > but then that means we're possibly going to fail here. Example:
>
> [snip]
>
> > I guess it would make sense to fix this for real instead of hiding it a
> > bit further. Unfortunately 4am isn't a great time to set up a reproducer
> > and to keep on hacking. :/
>
> There are in effect two bugs here. The first, which the patch fixes,
> is that any folder with spaces in its name will cause iso-scan's
> postinst to fail, preventing the installation. The second, which the
> patch doesn't fix, is that any ISO found in a folder with spaces in
> its name won't be handled correctly.
>
> The first bug is extremely confusing, since an otherwise OK USB key
> with all the appropriate files in the right place will fail to
> install, with no explicit error message, and worse than that with a
> message on the fourth terminal indicating that the ISO was found and
> is usable...
>
> The second bug, which is mitigated in part by the existence test (line
> 170), will only prevent certain ISOs from being used, and won't abort
> the installation.
>
> Wouldn't it be acceptable to apply the patch, and add an erratum
> indicating that ISOs shouldn't be placed in folders containing spaces
> in their names?
Thanks for your analysis. If the consequences are those you mention I
agree that fixing the first bug right away (backporting as needed) is
desirable.
I'll probably clone this bug report for the second one, and deal with
the first one in a moment.
Mraw,
KiBi.
Attachment:
signature.asc
Description: Digital signature