Dear ASDF users,
can you please test ASDF 2.017.22? I'm declaring a feature freeze for ASDF until release, and assuming no bug is found and bugfix committed, the current ASDF will be promoted as 2.018 by the end of the month. The repository is still open for documentation and test updates.
On the other hand, for any extensions you might want to publish, (for instance, around hooks for package renamings), I've created a asdf-contrib.git repo, currently empty save for a trival .asd file, next to the asdf.git repo: ssh://common-lisp.net/project/asdf/git/asdf-contrib.git git://common-lisp.net/projects/asdf/asdf-contrib.git
Changes since 2.017: * Solved a bug due to ASDF reusing components even when the ASDF definition has changed wildly, leading to bad behavior in incremental compilation and sometimes total failure of compilation of a new version. (Thanks to Xach for reporting and RPGoldman for fixing.) * Bind *default-pathname-defaults* around load-sysdef, to increase chances of successful pathname parsing in defsystem. * Many implementation specific improvements for ECL: merge with the ECL upstream version, support for the bytecode compiler, * Fix an issue with ASDF doing the wrong thing in case of a clock skew, thanks to Andreas Fuchs. * Simplification of the internals of perform-with-restarts and traverse's do-dep, that should fix some corner cases and make it easier to extend or further fix. There scarily doesn't remain any unedited code from Dan Barlow's original ASDF, though much of the interface design is preserved. * Improvements to tests and documentation. * Upgrade code slightly simplified, with a particular fix for CLISP; it is now tested on all supported implementations, not just SBCL. * Fixed run-shell-command on allegro/unix and clisp, documented the function, noted how it mightn't do what you think on Windows, marked it as obsolete, promoted as an official replacement my new xcvb-driver:run-program/process-output-stream * New :around-compile hook so you can control the syntax, optimization, and warning level, rename packages and balance compile-time side-effects, etc., around the compilation of files in your own modules. * Because it is required at least on ABCL, made the Unix vs Windows detection a runtime rather than compile-time distinction No more asdf-unix or asdf-windows feature, but (asdf::os-unix-p) and (asdf::os-windows-p) functions (not exported for now). Also recognize darwin as Unix ECL.
—♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org No man would listen to you talk if he didn't know it was his turn next. — Edgar Watson Howe