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

Re: The horrible infinite-loop detected bug

On Thu, 17 Dec 2009, Frank Küster wrote:
> it looks okay up to this point:
> >       *)
> >         if [ $disable_from_now = 1 ] ; then
> >           echo "# $line" >> $tmpfile
> >         else
> >           echo "$line" >> $tmpfile
> >         fi
> >         ;;
> >     esac
> This looks dangerous to me:  If someone has added local custom formats
> (and not followed the advice of doing this in a separate configuration
> file), then they'll probably have done this at the very end.

Yes. I agree. We should also check that the following lines actually
define the two formats as taken over from the old file. What a pain.

In addition, the code as it is now will jump in every time we upgrade
and add '#' to all the lines again and again and again and again which
is sub-optimal.

> Am I right that in TL 2007 the duplicate lines were no problem?  Then
> such systems might well exist.

I think so, AFAIR the fmtutil-sys loop check was added later. In fact
it is a bug in fmtutil, too, because it should reset the code if it
starts with a new format entry. I will take a look into that, too, but
we *SHOULD* get rid of the old config files entries in any case.

But, in fact people having this problem MUST have rejected upgrade of
conffiles, am I right? Otherwise it would not break? Isn't there a
way we can force them to ugrade conf files if it breaks the system
NOT to do it???

> We could at least stop at blank lines.  Or restrict the code like this:
>       etex*|pdfetex*)
>         if [ $disable_from_now = 1 ] ; then
>           echo "# $line" >> $tmpfile
>         else
>           echo "$line" >> $tmpfile
>         fi
>         ;;
>       *)
>         echo "$line" >> $tmpfile
>         ;;

Yes, something looks reasonable.

>   echo "Some formats have been disabled in $oldfile" >&2
>   echo "Please review the changes!" >&2
> That will surely give you bad comments from debconf advocates - and I
> think they at least have the project's majority (or louder voices)
> behind them, if not even policy.

??? I am not allowed to *interact* with the user, but I might send
out stuff to the log ... we do it permanently ...
	this may take a while .... done.

Best wishes


Norbert Preining                preining@{jaist.ac.jp, logic.at, debian.org}
JAIST, Japan               TU Wien, Austria            Debian TeX Task Force
gpg DSA: 0x09C5B094   fp: 14DF 2E6C 0307 BE6D AD76  A9C0 D2BF 4AA3 09C5 B094
An extremely intricate knot originally used for belaying the
topgallant foresheets of a gaff-rigged China clipper, and now more
commonly observed when trying to get an old kite out of the cupboard
under the stairs.
			--- Douglas Adams, The Meaning of Liff

Reply to: