From strandh at labri.fr Fri Dec 8 06:51:56 2006 From: strandh at labri.fr (Robert Strandh) Date: Fri, 8 Dec 2006 07:51:56 +0100 Subject: [stamp-devel] what language? Message-ID: <17785.2956.526233.676087@serveur5.labri.fr> Hello, Since this list is archived and since there is likely to be people here in the future who do not understand French, I suggest we use English for this list. OK? -- Robert Strandh --------------------------------------------------------------------- Greenspun's Tenth Rule of Programming: any sufficiently complicated C or Fortran program contains an ad hoc informally-specified bug-ridden slow implementation of half of Common Lisp. --------------------------------------------------------------------- From strandh at labri.fr Sun Dec 31 06:05:38 2006 From: strandh at labri.fr (Robert Strandh) Date: Sun, 31 Dec 2006 07:05:38 +0100 Subject: [stamp-devel] Some initial thoughts Message-ID: <17815.21298.484324.673931@serveur5.labri.fr> 1) I have checked in stamp.lisp and a very preliminary specification in Spec/spec.text 2) Using mel-base is going to create some problems. For one thing, mes-base seems to use base-string and base-char a lot, but that seems to be incompatible with the kind of strings used by SBCL. The solution would be to fix the declarations in mel-base. Perhaps this has already been fixed; I tripped on this problem by using an old version of mel-base. Another problem with mel-base is that it can't handle non-ascii contents of message headers and message bodies. The solution to this problem seems to be to make mel-base use flexi-streams, to scan the header in ascii mode with a default character (as flexi-streams allow) of (say) #\?, and to set the encoding of the stream for the body to whatever the content-type/charset header indicates (with a default of ascii as for the header). This would require some digging into the mel-base code, but I think it can be done. Mel-base is pretty well written. To get around this problem until mel-base is fixed, I suggest using a test folder that has only ascii characters in all of the messages. We don't want to use the `messages' function on a folder, simply because it scans the entire directory and we might have a million files in there. Instead, we must select an individual message based on the tags list, and let mel-base scan that individual message. 3) I am working on a list of priorities. I think I'll post it here and leave it open to debate. In a week or so, we should have nailed the priorities so that the students know what to work on and in which order. 4) I still intend to convert the existing Stamp code to use the ESA library, but I don't know at which speed this will happen. It would be good to have something running in a week or so. 5) We need to think about file and package structures. We should definitely create an ASDF file (named `stamp.asd'), and a `packages.lisp' file. The main package should be called `stamp', and should contain only exported symbols. Other packages used for implementation should :use the stamp package (I learned this trick from CLIM). We need to decide what other packages are required. For now, let's say we only need one: stamp-internals, perhaps with a nick name of `stampi'. We should probably have this working in a week or so as well. Comments? -- Robert Strandh --------------------------------------------------------------------- Greenspun's Tenth Rule of Programming: any sufficiently complicated C or Fortran program contains an ad hoc informally-specified bug-ridden slow implementation of half of Common Lisp. --------------------------------------------------------------------- From strandh at labri.fr Sun Dec 31 06:25:17 2006 From: strandh at labri.fr (Robert Strandh) Date: Sun, 31 Dec 2006 07:25:17 +0100 Subject: [stamp-devel] priorities Message-ID: <17815.22477.849059.749734@serveur5.labri.fr> Actually, I think I'll create a `priorities.text' file in the `Spec' directory instead of posting priorities here. It will be much easier to manipulate that way. -- Robert Strandh --------------------------------------------------------------------- Greenspun's Tenth Rule of Programming: any sufficiently complicated C or Fortran program contains an ad hoc informally-specified bug-ridden slow implementation of half of Common Lisp. --------------------------------------------------------------------- From strandh at labri.fr Sun Dec 31 08:04:16 2006 From: strandh at labri.fr (Robert Strandh) Date: Sun, 31 Dec 2006 09:04:16 +0100 Subject: [stamp-devel] why are CVS commit messages not forwarded to stamp-cvs? Message-ID: <17815.28416.602654.718443@serveur5.labri.fr> Is that a configuration thing? -- Robert Strandh --------------------------------------------------------------------- Greenspun's Tenth Rule of Programming: any sufficiently complicated C or Fortran program contains an ad hoc informally-specified bug-ridden slow implementation of half of Common Lisp. ---------------------------------------------------------------------