I have recompiled mod_lisp2.c against Apache 2.2.8 for Windows using Visual C++ Express 2008. mod_lisp itself only needs the include files and libraries that are an optional part of the Apache binary installation, but I downloaded the Apache source (that includes Visual Studio 6 project files) so that I could check the compiler and linker settings.

My slightly modified mod_lisp2.c, the Visual C++ 9 project file and the Release build of the module are available in a .zip file. Please note the title of this blog when considering whether to use any of it.

The diff looks like this:

*** ../../download/programming/Other Lisp stuff/mod_lisp2 - svn.c Tue Apr  8 23:33:07 2008
--- mod_lisp2.c Mon Jun  2 18:15:31 2008
***************
*** 443,451 ****
  {
    char crlf[2] =  {0xd, 0xa};
    char length[16];
!   snprintf(length, 16, "%x", n_bytes);
    
!   apr_status_t status = write_lisp_data (socket, length, strlen(length));
    if ( status == APR_SUCCESS)
      {
        status = write_lisp_data (socket, crlf, 2);
--- 443,453 ----
  {
    char crlf[2] =  {0xd, 0xa};
    char length[16];
!   apr_status_t status;
! 
!   apr_snprintf(length, 16, "%x", n_bytes);    // 2 Jun 08 JDP
    
!   status = write_lisp_data (socket, length, strlen(length));
    if ( status == APR_SUCCESS)
      {
        status = write_lisp_data (socket, crlf, 2);

If there were one Common Lisp which I could run on every platform — one which provided portability for applications using threads and doodads and whatsits without requiring me to GPL my code because I use some internal interface of that implementation — I would probably use it, even if it provided less performance than my current lisp of choice. When it comes down to it, I rarely need performance. No such flower or weed exists yet. If I were to plant it, I would not plant it in the garden of Common Lisp. For as much as I love working in this language I would not take the opportunity to implement any of the obvious advances in language design in the past twenty years and pass it by just to implement ANSI Common Lisp.

And then I think: why don't I just buy a couple of LispWorks licenses for my application development needs, and then use Factor as the FOSS, cross-platform, natively-compiled, metaprogramming-capable, embeddable language for Ngake? At least until Arc is ready, although that could be many years away...

Meanwhile, the CL-Gardeners list continues to throw up links to useful Lisp projects, like Chris Riesbeck's Lisp Critic.

Updated 21/12/05: Alistair has posted more information to the CL-Gardeners list. ]]> Mon, 19 Dec 2005 23:06 GMT lisp/some-lisp-links http://johnp.net/blog/lisp/some-lisp-links.html

• Kiwi, a CLiki replacement.
• Gardeners, a mailing list started by Peter Siebel. Subtitled "Tending the Lisp Garden." He has posted some notes on the list's purpose. The listinfo page is here. I joined and posted a list of my intended projects for Corman Lisp.
• Duncan Rose, porting DUIM to Lisp. (Via the CL Gardeners list.)
• RT, a Lisp regression testing framework (selected by Duncan).

1. Set up a patches page listing all the sets of patches people have prepared for Corman Lisp:
   • Chris Double's patches (if there are any that haven't already been folded into the official distribution)
   • Edi Weitz's ASDF patches
   • Espen Wiborg's SLIME patches
   • And of course any patches I manage to come up with...
   It occurred to me that a darcs repository with current versions of all these patches could be set up, and then zip files of modified sources containing any desired combination of patches could be automagically generated.
2. Install Paul Dietz's GCL ANSI test suite and run it against Corman Lisp.
3. Try to run Pascal Costanza's CLOS feature tests against Corman Lisp.
4. Try to port Pascal's Closer to MOP compatibility layer to Corman Lisp.
5. Install CFFI and Verrazano. Try to run them against the Torque Game Engine. Compare them to something based on the ideas behind FuBi.
6. Get Corman Lisp to compile under GCC/MinGW, verified by the GCL ANSI test suite.

Reading the Fetter/VZN mailing list archives was interesting. It sounds like its at a pretty workable state. They don't seem to have found GCCXML to be a limiting factor, and their handling of overloaded names is, er, maximally straightforward. ]]> Thu, 13 Oct 2005 08:34 GMT lisp/lisp-libs-and-continuations http://johnp.net/blog/lisp/lisp-libs-and-continuations.html Chris Double links to a paper by Joe Marshall (whose FOG code I've linked to already) about implementing first-class continuations in languages like Java or C#. Must read.

And Gary King has finally posted about Lisp, and actually links to several libraries he's preparing for release. I often think that I should study people's Lisp utility libraries, as I'm sure that would be educational.

On the SLIME mailing list, someone linked to this new evaluator for SBCL. ]]> Mon, 10 Oct 2005 09:50 GMT lisp/closer-project http://johnp.net/blog/lisp/closer-project.html Chris Double's blog, the Closer project has a suite of MOP tests and a MOP compatibility layer designed to provide a consistent foundation for heavy MOP work across different Lisp implementations. Only tested on OS X so far by the looks of it, but it'd be amusing (OK, maybe educational) to throw it at Corman Lisp and see how far it gets. ]]> Thu, 25 Aug 2005 14:27 GMT lisp/glr-parsers-in-lisp http://johnp.net/blog/lisp/glr-parsers-in-lisp.html post by Dan Knapp (who also has a useful-looking guide to proxying a Lisp web server with Apache). He links to an interesting-looking report on scannerless GLR parsing, which I shall print out and attempt to read. There's also dparser, a scannerless GLR parser written in C, which has a manual and looks quite complete. ]]> Thu, 11 Aug 2005 00:41 GMT