[Prev][Next][Index][Thread]
Re: CMU-CL-HTTP
|From: "Harper, Michael L." <Michael.Harper@alcoa.com>
|To: "'Conrad Bookout'" <conrad@infotank.com>,
| "'Paul F. Werkowski'"
| <pw@snoopy.mv.com>
|Cc: "'www-cl@ai.mit.edu'" <www-cl@ai.mit.edu>
|Subject: RE: CMU-CL-HTTP
|Date: Wed, 27 Nov 1996 16:09:39 -0500
|X-Mailer: Microsoft Exchange Server Internet Mail Connector Version 4.0.994.57
|Encoding: 78 TEXT
|
|I have two items to mention. If you are running on x86 Linux, Franz
|plans to release a free port of its current ACL 4.3 system for
|non-commercial Linux use next month.
|
|Second, to Mr. Werkowski, what are you referring to in your last
|>paragraph below when you say "There is also the misuse of the values
|declaration throughout which causes Python problems."????
|
|Mike Harper
|
Sorry, If I was aware my reply was going to the list I would
have phrased that differently. This issue has been brought up
(by me) before.
>From a CMUCL-centric point of view, a values declaration wants
to have the TYPE of return values mentioned in the declaration
list, eg (declare (values list float foo)). The values declaration
used by CL-HTTP implementors seems to follow another drummer
which seems to put variable names instead of TYPES. This causes
CMUCL's type-aware Python compiler to bitch like hell.
Outside of changing all the source code, a CMUCL port
needs to put something like this in the sysdecl file.
#+cmu17 ;; Don't warn about botched values decls
(setf c:*suppress-values-declaration* t)
which stops the complaints. This is OK I guess since the
values variable-name declarations are (in my view) meaningless
anyhow.
Paul
References: