Guile bug could affect Windows users

[Charles Day]

On Jan 23, 2008 6:25 AM, Derek Atkins <warlord at mit.edu> wrote:

> Quoting Mike or Penny Novack <stepbystepfarm at mtdata.com>:
>
> >
> >>
> >> Additionally, when building the setup.exe on Windows we compile guile
> from
> >> source anyway, so applying a patch that has been sent to bug-guile but
> is
> >> not contained in any release yet is perfectly fine, at least IMHO.
> >>
> >> -- andi5
> >>
> >>
> >>
> > Again I am probably making some wrong assumptions. Compiling WHAT? The
> > interpreter certainly, "eval" itself, perhaps a few of the most
> > basic/most frequently used function definitions. But with most
> > "extendable" languages like LISP and its dialects the bulk of the
> > definitions available at the start are usually done in the language
> > itself. OR, in "compile and go" implementations that makes little
> > difference (any that are added are compiled before execution/evaluation
> > as opposed to a strict "interpreter" implementation where the expression
> > would be reinterpreted every time used.)
> >
> > Isn't that true of Guile? Some reason to believe that these particular
> > erroneous definitions were part of the compile? Even if they were, what
> > difference would that make except perhaps speed when being evaluated?
> > They still could be overridden by redefinition, yes?
> >
> > (I hope not confusing people with bringing up matters of how LISP like
> > languages are implemented..)
>
> I think what he meant is that we start with the guild source tree
> and create our own guile environment from source.  In other words, we
> control the guile environment on windows so we can apply this patch.
> Whereas on other platforms we assume the OS/Distro provides guile
> to us and we just use it.
>

The bug is in the Guile source code, in libguile/filesys.c.  In truth, I
made the (perhaps incorrect) assumption that the GnuCash project would not
want to maintain custom fixes to Guile. If that's an option, a more detailed
description of the bug, along with my very simple patch can be found
here<http://lists.gnu.org/archive/html/bug-guile/2008-01/msg00041.html>
.

But as I said, this bug is not really a big deal at the moment because
GnuCash isn't actually using the dirname or basename procedures. (I grep'd
all the .scm files and found no matches.)  So I just wanted to warn anyone
who might decide to begin using these procedures.

-Charles



>
> -derek
> --
>       Derek Atkins, SB '93 MIT EE, SM '95 MIT Media Laboratory
>       Member, MIT Student Information Processing Board  (SIPB)
>       URL: http://web.mit.edu/warlord/    PP-ASEL-IA     N1NWH
>       warlord at MIT.EDU                        PGP key available
>
> _______________________________________________
> gnucash-devel mailing list
> gnucash-devel at gnucash.org
> https://lists.gnucash.org/mailman/listinfo/gnucash-devel
>