- 29 Oct, 2016 2 commits
-
-
Robert Goldman authored
-
Robert Goldman authored
Check for and reject attempts to directly invoke MAKE-INSTANCE on OPERATION classes. Add a test for the error condition. Squashed commit of the ENFORCE-MAKE-OPERATION branch.
-
- 26 Oct, 2016 6 commits
-
-
Robert Goldman authored
uiop/pathname: remove export of non-existent func The function was removed in 039494bc. See merge request !42
-
Robert Goldman authored
Master+release Merge release back into master, to resolve divergence and avoid future issues. See merge request !41
-
Robert Goldman authored
-
Robert Goldman authored
No changes to code, just some notes for the reader.
-
Robert Goldman authored
Fix run-program's file existence check interface This fixes https://bugs.launchpad.net/asdf/+bug/1629680. See merge request !35
-
- 25 Oct, 2016 1 commit
-
-
Francois-Rene Rideau authored
-
- 19 Oct, 2016 1 commit
-
-
Robert Goldman authored
Fix release symlinks to make Quicklisp happy I investigated why Quicklisp still hadn't picked up release 3.1.7 or uiop. It looks like the Makefile was creating symlinks in the wrong directory. See merge request !39
-
- 18 Oct, 2016 3 commits
-
-
Francois-Rene Rideau authored
I investigated why Quicklisp still hadn't picked up release 3.1.7 or uiop. It looks like the Makefile was creating symlinks in the wrong directory. Update index.html to link to the symlinks in archives/
-
Robert Goldman authored
Copied updates over from master (3.1.7.30), because archive tarballs were not constructed correctly: archives didn't have prefix directories; version.lisp-expr was missing; tarballs were not properly named.
-
Robert Goldman authored
The tar files were not getting proper prefix directories. There were some mistakes in naming. version.lisp-expr was missing from the UIOP tarball.
-
- 17 Oct, 2016 1 commit
-
-
Robert Goldman authored
UIOP was not shipped with version.lisp-expr. There was a redundant, no-op mv in the makefile rule for "archive."
-
- 16 Oct, 2016 10 commits
-
-
Robert Goldman authored
-
Robert Goldman authored
Stop fmakunbound'ing defgeneric's past the upgrade trigger Avoid using `fmakunbound` unless necessary, and clear registered systems if we do. Audit existing gf's and conclude that only `component-depends-on`, `input-files` and `find-component` need it, and that the `*oldest-forward-compatible-version*` is 3.1.7.20. This is a response to https://bugs.launchpad.net/asdf/+bug/1631771 See merge request !36
-
Robert Goldman authored
-
Robert Goldman authored
No longer needed. Only needed to fix upgrade from version 2.x, but there's no more 2.x on Allegro.
-
Robert Goldman authored
Reorganized the package-moving logic. Previously it mingled lines of code for allegro/non-allegro. Now the Allegro and non-Allegro lines are grouped together. This made it easier to deal with a LET binding unused on Allegro.
-
Francois-Rene Rideau authored
Apparently, rename-package is badly broken on Allegro.
-
Francois-Rene Rideau authored
Get rid of undefined-function and any clisp or ecl specific code in it in uiop/utility as well as of a lingering #-clisp in asdf/upgrade. Whatever issues previous generations of upgrade code tried to address, they happily do not apply today. Update test-sysdef-asdf to better test asdf self-upgrade.
-
Francois-Rene Rideau authored
We just re-defined the signature of find-component and it needs to be fmakunbound. component-depends-on and input-files have some :around methods removed at 3.1.2 and it's safer to fmakunbound them if inherited from earlier versions. All other functions seem A-OK since 2.27, and we punt on 2.26 and earlier. See discussion in asdf/asdf!36
-
Francois-Rene Rideau authored
-
Francois-Rene Rideau authored
If we fmakunbound a defgeneric, we lose any user-defined method on the function and must therefore reload any and every system that might have defined methods on it. We must therefore not fmakunbound past the forward compatibility upgrade trigger, only before. Adjust our code accordingly. Conservative bounds are used in *oldest-forward-compatible-asdf-version* and redefined-functions that should probably be refined after some inspection of old versions. This is a response to https://bugs.launchpad.net/asdf/+bug/1631771
-
- 14 Oct, 2016 1 commit
-
-
Francois-Rene Rideau authored
Make sure that bundle-op, concatenate-source-op, Makefile, make-asdf.sh, make-asdf.bat and tools/release.lisp all agree on "the" proper order of files in uiop and asdf.
-
- 12 Oct, 2016 3 commits
-
-
Elias Pipping authored
Add a test, too.
-
Elias Pipping authored
Add a test, too.
-
Elias Pipping authored
Add a test, too.
-
- 09 Oct, 2016 2 commits
-
-
Robert Goldman authored
-
Francois-Rene Rideau authored
This avoids a forward reference on sbcl/windows. Also add a comment on a use of symbol-call on ECL.
-
- 08 Oct, 2016 9 commits
-
-
Elias Pipping authored
-
Francois-Rene Rideau authored
DIRECTORY is important. It was never actually ignored, except on Allegro, on which the implementation itself handles the option. Make that explicit. The other arguments are always passed... though it's unclear whether they are heeded in all implementations, since we have :allow-other-keys t. That deserves investigation and a comment update, at some point. Also, tweak some comments.
-
Francois-Rene Rideau authored
Use explicit error checking instead.
-
Francois-Rene Rideau authored
Thanks to Elias Pipping for his help getting this commit right.
-
Francois-Rene Rideau authored
-
Francois-Rene Rideau authored
Make %run-program the alias of launch-program rather than the other way around. Rename %use-run-program to %use-launch-program. Update tests. NB: I wanted to move %run-program to backward-driver, but that won't help: since the function was never exported from uiop/run-program, it wasn't in uiop. Thanks to Elias Pipping for his help getting this commit right.
-
Elias Pipping authored
The run-program interface generally behaves as if :if-input-does-not-exist :error :if-output-exists :supersede were specified and deviations from this set will generally not be handled correctly, e.g. on CLISP or whenever %system is used. Hence we remove the corresponding tests for now
-
Elias Pipping authored
If we call it synchronously, we also need to call wait-process and close-streams. We ignore the error status because signalling a dead or non-existing process is not something that the user can avoid.
-
Elias Pipping authored
-
- 06 Oct, 2016 1 commit
-
-
Robert Goldman authored
-