From fahree at gmail.com Sat Apr 3 18:44:29 2010 From: fahree at gmail.com (=?ISO-8859-1?Q?Far=E9?=) Date: Sat, 3 Apr 2010 19:44:29 +0100 Subject: [xcvb-devel] XCVB .494, now with CL_SOURCE_REGISTRY Message-ID: I haven't been publicly committing for two months, because I started a refactoring and broke everything, and it took me some time to clean things up. The refactoring is about having a layer of "virtual-pathnames" between the graph model and the actual pathnames. I would have used logical pathnames if they weren't completely non-portable, with SBCL strictly enforcing the CLHS restrictions. It was just a preparation for future changes, but it was painful. Other minor bugs have been fixed. Another big change is that XCVB now relies on the latest ASDF 1.663 and associated cl-launch 2.902, and shares the same new mechanism for locating source trees. Goodbye XCVB_PATH, hello CL_SOURCE_REGISTRY! It passes the regression test. Yay for regression testing! [ Fran?ois-Ren? ?VB Rideau | Reflection&Cybernethics | http://fare.tunes.org ] If you could make a programming language where programmers could write in english, you'd find that most people can't write english. ? "Masoud Pirnazar" on comp.lang.lisp 2001-10-12 From fahree at gmail.com Wed Apr 14 08:38:49 2010 From: fahree at gmail.com (=?ISO-8859-1?Q?Far=E9?=) Date: Wed, 14 Apr 2010 04:38:49 -0400 Subject: [xcvb-devel] XCVB .500, with fewer bugs Message-ID: .500 passes the regression test suite, that has been augmented to include a generated file in the mix. It also resolves module names according to spec, which wasn't the case before in odd cases. It also knows about builds that are invalid because they inherit from a naming conflict. With .500, I declare XCVB back on track wrt development. There's still a long road ahead, but all the basic functionality is here again, more robust than before, and preliminary work has been done for further features. [ Fran?ois-Ren? ?VB Rideau | Reflection&Cybernethics | http://fare.tunes.org ] Fraud is the homage that force pays to reason. ? Charles Curtis From fahree at gmail.com Thu Apr 29 21:43:11 2010 From: fahree at gmail.com (=?ISO-8859-1?Q?Far=E9?=) Date: Thu, 29 Apr 2010 17:43:11 -0400 Subject: [xcvb-devel] XCVB .510 - stable again and with bug fixes Message-ID: XCVB .510 is stable again. Passes all self-tests on sbcl, ccl, clisp, though I had to disable POIU with CLISP, because it somehow fails (looks like CLISP might be eating our SIGCHLD or something). Notable bug fixes include proper treatment of dependencies for generated files (oops), proper use of environment variables to specify a custom lisp binary, especially useful for XCCL=/path/to/single-threaded-ccl, fix a breakage that disabled CFASLs, etc. Also included in the release are the latest versions of ASDF and POIU, so it all works. Various internals were updated that will make some of the future features possible, with respect to virtualizing pathnames between the two passes of execution (graph building and computation walking). Speaking of which, if you have pointers on ways to better specify what effects may or may not happen when (where depending on the strategy, "when" may change), I'm curious. [ Fran?ois-Ren? ?VB Rideau | Reflection&Cybernethics | http://fare.tunes.org ] Judge and party ? the ultimate nature of a monopoly government.