[Gcl-devel] Re: gclcvs-2.7.0 bugs about macro-function and symbol-macrolet

[Camm Maguire]

Greetings!

"Hoehle, Joerg-Cyril" <address@hidden> writes:

> Camm,
> 
> as I currently maintain J.Amsterdam's Iterate package code base and have a 
> nice Ubuntu/Hoary (Debian based) Linux/686 box which makes it easy to install 
> pre-compiled software, I tried Iterate with gcl-cvs.
> 

Great!

> [I deleted gcl-cvs own Iterate/Gathering package for testing this, because it 
> has the exact same name and nickname, yet its unrelated, and I figured out 
> this would be faster than my renaming Amsterdam's Iterate]
> 

OK, but be careful -- pcl uses this to my understanding.

> Amsterdam's Iterate macro package is also quite old (but not as old as AKCL, 
> which I rarely used in the late 80ies). I'll have a look again at the 
> gathering package which I had long forgotten.
> 
> 
> Version gclcvs 2.7.0 ANSI Dec 12 2004 (2.7.0-36 in apt)
> 
> o GCL-CVS ANSI complains about 2 items:
> 
> - (macro-function x env): does not accept the second argument
>  -- probably causes several testcases to fail, because macroexpand needs to 
> get and pass around the environment.
> 
> - (symbol-macrolet ((x (* i i)))
>           (declare (optimize debug))
>           (sum x))
> "Encountered declare... in a place where a declare was not expected"
> 

Thank you so much for your report!  As you know, ansi is still a work
in progress for gcl.  It would be most helpful if you could double
check that the specific errors you see are covered in the existing
failures reported in Paul's ansi test suite.  They are sure to get
fixed if there.  If not, consider asking Paul to add such a test.  If
this does not work, it would be most helpful if you could register
these as bugs on the website so they do not get lost -- if this is
difficult, I'll be happy to do so in your behalf.

> 
> Interestingly, it also revealed a problem in one Iterate testcase, where a 
> function object wants to get compiled (instead of its lambda-expression), 
> which gcl doesn't grok but all other implementations do (gcl complains about 
> lisp:lambda-closure not being defined). I'll rewrite that section of Iterate, 
> because the function need not be turned into a fn object prior to 
> compilation, let the compiler do that -- just to illustrate that it's 
> worthwhile to test plenty of CL systems :-)
> 

A specific bug report here would also be most helpful if gcl is afoul
of ansi here.

Take care,

> Regards,
>       Jörg Höhle.
> 
> 
> 

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