gcl-devel
[Top][All Lists]
Advanced

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

Re: [Gcl-devel] On the package tests


From: Camm Maguire
Subject: Re: [Gcl-devel] On the package tests
Date: 09 Oct 2002 00:16:58 -0400

Greetings!  OK I've just committed some changes.  I'm now down to 172
out of 4148 failures.  

I think the biggest item that remains with the package stuff is the
error handling.  For example, when I read the spec for delete-name, it
appeared to state that a continuable package-error should be triggered
if the package is used by others or does not exist.  When I
implemented this, I caused several additional test failures.  Also, I
can't find where the function package-error is defined in the spec
either.

I'm hoping that the field is small enough
now that you might be able to shed some additional light as to how the
remaining failures should work.  

Remaining cleanups include implementing package documentation strings,
and to check the .lsp code for instances of in-package which need to
be changed.  This latter was perhaps the most significant of the
modifications -- the old gcl implementation of in-package would call
make-package if the designator did not exist.  Many of the init
scripts relied on this to setup their packages.  I've replaced the
relevant instances in init_gcl.lsp.in and pcl/sys-package.lisp.

Lastly, I've discovered that our compiler variable setup was turning
on optimization at all times regardless of the *speed* setting.  I've
added the additional configure variables O3FLAGS and O2FLAGS to allow
setting these at compile time.  I didn't modify the Windows section in
cmpmain.lsp to use the new variables, so if the Windows maintainers
would like to try this out, please do.

Take care,

"Paul F. Dietz" <address@hidden> writes:

> Comments on the tests:
> 
> (1) Many package functions take package designators as an argument.
>    A package designator can be a package object or a string designator.
>    A string designator can be a string, a character, or a symbol
>    (or other things on which the function STRING does not fail.)
>    Many tests are failing because string designators are not handled,
>    or not handled fully.
> 
> (2) gcl doesn't implement the DELETE-PACKAGE function, even in ansi
>    mode.  Many tests set up packages, do something, then clean up
>    after themselves with DELETE-PACKAGE.  They're all failing now.
> 
> I found one test (FIND-SYMBOL.8) that was busted; it's been fixed.
> 
>       Paul
> 
> 
> 
> _______________________________________________
> Gcl-devel mailing list
> address@hidden
> http://mail.gnu.org/mailman/listinfo/gcl-devel
> 
> 

-- 
Camm Maguire                                            address@hidden
==========================================================================
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah




reply via email to

[Prev in Thread] Current Thread [Next in Thread]