On Mon, 11 Apr 2005, Rankin, SE (Stephen) wrote:
> All,
>
> Also, forgot to say that the size_test.sh fails:
>
> PASS: scb_test.sh
> input buffer overflow, can't enlarge buffer because scanner uses REJECT
> FAIL: size_test.sh
> ===================================
> 1 of 2 tests failed
> Please report to [log in to unmask]
> ===================================
Hi Steve,
This is known; I discussed it in an email of 28 Feb 2005:
> 2. Since SCB isn't that much use at the moment (it can only index
> a source archive in mk-type format) just get rid of the -Sflex.skl
> from configure.ac and leave a note to the effect that this has
> been done. It will still work under these circumstances,
> it'll just miss routines in some of the files.
>
> I've spent a while trying to do something which has the same effect
> as this skel file trick without using -S, but with no success
> (evidently I didn't when I was first writing it either). It probably
> is possible, but I'm reluctant to spend much more time on it, since
> it's not really high-priority work. So if you leave it to me, I'll
> do (2) above.
>
> What I have done is to introduce a new test (make check target)
> which will test whether the version you've just built chokes on
> big AST files or not.
Failing is the 'right' thing to do, since SCB is known to be somewhat
(though not disastrously) broken on some systems in this respect,
and as mentioned above I'm not inclined to work hard to fix it just now.
Does it cause problems at some level for the build system that this
happens? If so I'll think of something to do about it (like delete
the test).
As to whether test_size.sh gets into the dist file or not, I can't
throw any light on that, though EXTRA_DISTing it sounds eminently
sensible.
Mark
--
Mark Taylor Starlink Programmer Physics, Bristol University, UK
[log in to unmask] +44-117-928-8776 http://www.star.bris.ac.uk/~mbt/
|