Andreas Fuchs wrote:

On 2006-03-12, Andreas Fuchs asf@boinkor.net wrote:

...

This mail marks the start of the freeze period for McCLIM 0.9.2.

Erm, not quite. Last-minute feature greediness on my part (and Tim's promise to get drag & drop of presentations to work and to not break anything) have made me reconsider starting the freeze today.

An updated and final freeze mark mail will go out within 24 hours, when (hopefully) Tim has gotten presentation drag translators to work or when we backed out the change.

Thanks for your patience, (:

I want to thank the release engineer's indulgence, however selfish his motivations :) I think drag-and-drop is in good enough shape to be labeled as "experimental" and not disrupt the rest of the system. I may try to sneak in a fix or two for the highlighting in drag-and-drop, but I promise that nothing will be major.

One thing I think worth hacking on even during the freeze is the Beagle back end, which doesn't work right now due to bit rot. A couple of methods need to be implemented to at least get the address book demo working again. It would be really nice to ship the release with the Beagle back end at least limping along.

We don't have a good bug data base other than Paolo's wiki page at http://mcclim.cliki.net/Bug. This is unfortunate and should be addressed after the release in order to get ready for the next one (in a couple of months, right Andreas?), but it's all we've got for now. I'm going to add another one soon. We should take a look and see if any can be fixed easily now and which ones should be a priority for next release.

Tim

Andreas Fuchs asf@boinkor.net writes:

This mail marks the real start of the freeze period for 0.9.2. Please be conservative in choosing what you commit, and aggressive in choosing what you test.

Well, you asked for it. I have tried building a freshly checked out copy of McCLIM's CVS repository with CMUCL Snapshot 2005-11 (19C) under Slackware Linux 10.0. Compilation aborts with this error:

[...] ; /home/paolo/src/dev/mcclim/Goatee/presentation-history.x86f written. ; Compilation finished in 0:00:00. ; Loading #P"/home/paolo/src/dev/mcclim/Goatee/presentation-history.x86f". ; Loading #P"/home/paolo/src/dev/mcclim/Backends/PostScript/sheet.x86f".

Execution of a form compiled with errors: (DEFMETHOD SHEET-DIRECT-MIRROR ((SHEET POSTSCRIPT-STREAM)) (POSTSCRIPT-STREAM-FILE-STREAM SHEET)) [Condition of type KERNEL:SIMPLE-PROGRAM-ERROR]

Restarts: 0: [CONTINUE] Return NIL from load of #P"/home/paolo/src/dev/mcclim/Backends/PostScript/sheet.x86f". 1: [RETRY ] Retry performing #<ASDF:LOAD-OP NIL {58BF9FC5}> on #<ASDF:CL-SOURCE-FILE "sheet" {58E57875}>. 2: [ACCEPT ] Continue, treating #<ASDF:LOAD-OP NIL {58BF9FC5}> on #<ASDF:CL-SOURCE-FILE "sheet" {58E57875}> as having been successful. 3: [ABORT ] Return to Top-Level.

Debug (type H for help)

(C::DO-CALL #<Code Object "Top-Level Form" {5A60E7F7}> 78 79 4 ...) Source: Error finding source: Error in function DEBUG::GET-FILE-TOP-LEVEL-FORM: Source file no longer exists: target:code/byte-interp.lisp. 0]

Paolo

Today, Paolo Amoroso amoroso@mclink.it wrote:

Paolo Amoroso amoroso@mclink.it writes:

...

Execution of a form compiled with errors: (DEFMETHOD SHEET-DIRECT-MIRROR ((SHEET POSTSCRIPT-STREAM)) (POSTSCRIPT-STREAM-FILE-STREAM SHEET)) [Condition of type KERNEL:SIMPLE-PROGRAM-ERROR]

I tried to build on the same Linux box the latest McCLIM CVS sources with CMUCL Snapshot 2006-02 (19C), which is the latest available. Snapshot 2006-03 binaries are unavailable for Linux, due to a bug that was still present when the sources on which the snapshots are built were tagged. McCLIM compilation aborts with the same error.

Right, I see it, too. Seems like cmucl 19c has problems with methods specializing on direct subclasses of forward-referenced classes. Here's a patch working around that problem with asdf dependencies. McCLIM builds & runs fine for me on sbcl and cmucl with this patch; I have no postscript code (besided Examples/postscript-test.lisp, which works fine) to test it, though.

Here's the patch:

Thanks for the report and the reminder,

On 2006-03-19, Paolo Amoroso amoroso@mclink.it wrote:

Andreas Fuchs asf@boinkor.net writes:

...

classes. Here's a patch working around that problem with asdf dependencies. McCLIM builds & runs fine for me on sbcl and cmucl with this patch; I have no postscript code (besided

With your patch McCLIM builds fine. When I start the CLIM Listener, however, it aborts with this error:

• (clim-listener:run-listener)

Error in KERNEL::UNDEFINED-SYMBOL-ERROR-HANDLER: the function LISP::DYNAMIC-SPACE-USAGE is undefined. [Condition of type UNDEFINED-FUNCTION]

Hm, running the listener works for me on a debian-installed 19c on x86. My mcclim cvs checkout has #+cmu(lisp::dynamic-usage), though, and no reference at all to dynamic-space-usage. Is that a modified tree? Are you testing on another platform?

or at least with the same name. Maybe the function was renamed to LISP::DYNAMIC-SPACE-SIZE.

No, that's the overall dynamic space size, not the used space. All that symbol guessing would be much easier if we had to guess exported symbols (-:

Cheers,

On 2006-03-20, Douglas Crosher dtc@scieneer.com wrote:

...

Hm, running the listener works for me on a debian-installed 19c on x86. My mcclim cvs checkout has #+cmu(lisp::dynamic-usage), though, and no reference at all to dynamic-space-usage. Is that a modified tree? Are you testing on another platform?

This change was erroneosly submitted by me in patches for SCL, sorry. The fix:

• Revert function 'dynamic-space-usage to 'dynamic-usage.

Thanks a lot for the fix, I applied it.