[Openmcl-devel] Clozure CL 1.9 released

Anton Vodonosov avodonosov at yandex.ru
Wed Mar 13 00:31:30 UTC 2013


08.03.2013, 22:59, "Gary Byers" <gb at clozure.com>:
> If this kind of testing had produced a bug report that said something like:
>
> "Given the following code (which is distilled from code in the HTML-ENTITIES
> project), calling INIT-HASHES in [certain versions of CCL on certain platforms]
> causes a spurious TYPE-ERROR to be signaled", then yes, I'd care about that
> a lot.
>
> In this particular case, it probably wouldn't have been too hard to generate
> such a bug report from the test results you posted.  (Knowing that "linux-x86"
> meant "32-bit Linux-x86" would also have been helpful.)  In other cases, it might
> have been much more difficult to identify and/or isolate a bug.

I think the cases where the bug may be tracked down will constitute the majority.

Yes, "linux-x86" means "32-bit Linux-x86". The lisp implementation name
"ccl-1.8-f95-linux-x86" is produced by asdf:implementation-identifier.

(LOAD langutils FAIL) means that (ql:quickload :langutils :verbose t) failed.

So, this report means: 

"In quicklisp 2013-02-17, (ql:quickload :langutils :verbose t) works for CCL 1.8,
but fails in CCL 1.9".

It doesn't automatically distill the relevant code from langutils library; but it
gives an easy way to reproduce the problem and specifies the interval of CCL
versions where the problem was introduced.

To further narrow the scope this testing could be done more often, for example
once a month. This is how I know what failures are caused by ASDF - by
performing independent tests of CCL 1.8 and CCL 1.8 + ASDF 2.32 and comparing
results.

Do you need more information regarding the html-entities and langutils failures?

Best regards,
- Anton



More information about the Openmcl-devel mailing list