MKCL 1.1.10 <http://common-lisp.net/project/mkcl/releases/mkcl-1.1.10.tar.gz> is now available for general use on the MKCL <http://common-lisp.net/project/mkcl> page along with the usual set of pre-compiled binaries for MS-Windows. This release is mostly a consolidation of all the bug fixes accumulated along the last 2 years and a half or so. Please consult the log on MKCL's gitlab repository <https://gitlab.common-lisp.net/mkcl/mkcl> for a detailed account of the changes. The content of the "contrib" directory has also been updated. Most notably, ASDF is now at 3.1.7 and CFFI at 0.18.0. Going back to MKCL 1.2.0 and its spectacular performance improvements both in time and space... Cheers, Jean-Claude Beaudoin
Dear Jean-Claude, did MKCL recently change the bundle file type from "fasb" to "fas"? ASDF seems to be failing its test-bundle.script now. For the record, I never liked the idea of .fasb; but if you make incompatible changes, please synchronize with me and other users. Also, it seems to me that si:mkcl-version and cl:lisp-implementation-version used to be the same (and/or the former didn't exist?), but now latter is more precise whereas the former is used for number .so. Is that correct? Is it OK if I support only the latest version of MKCL? Do you have an idea of how many users you have, and whether some insist on using old versions of MKCL? Finally, test-program.script now fails, —♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org On Wed, Jan 18, 2017 at 4:38 AM, Jean-Claude Beaudoin <jean.claude.beaudoin@gmail.com> wrote:
MKCL 1.1.10 is now available for general use on the MKCL page along with the usual set of pre-compiled binaries for MS-Windows. This release is mostly a consolidation of all the bug fixes accumulated along the last 2 years and a half or so. Please consult the log on MKCL's gitlab repository for a detailed account of the changes.
The content of the "contrib" directory has also been updated. Most notably, ASDF is now at 3.1.7 and CFFI at 0.18.0.
Going back to MKCL 1.2.0 and its spectacular performance improvements both in time and space...
Cheers,
Jean-Claude Beaudoin
On Thu, Feb 2, 2017 at 10:38 PM, Faré <fahree@gmail.com> wrote:
Dear Jean-Claude,
did MKCL recently change the bundle file type from "fasb" to "fas"? ASDF seems to be failing its test-bundle.script now.
No. Nothing has moved in a long time on that front.
For the record, I never liked the idea of .fasb; but if you make incompatible changes, please synchronize with me and other users.
That .fasb business is a legacy of ECL as you know from the history of asdf-bundle. I will duly inform you if I ever purposely consider some incompatible change in that area of MKCL.
Also, it seems to me that si:mkcl-version and cl:lisp-implementation-version used to be the same (and/or the former didn't exist?), but now latter is more precise whereas the former is used for number .so. Is that correct?
Yes, this was the first post-1.1.10 change to MKCL. This change should only affect git controlled development versions of MKCL. For released versions of MKCL the two functions are still identical. Under git, the preference is given to the new #'si::git-describe-this-mkcl instead, which gives a more exact description of the real version of the specific instance of MKCL. I thought that change was a good feature, maybe I was wrong.
Is it OK if I support only the latest version of MKCL?
I don't have any expectations in that area other than that latest ASDF release should work on latest MKCL master branch head or something close to this. Seeing ASDF be backward compatible with earlier MKCL version is surely a nice thing but it is an ASDF initiative.
Do you have an idea of how many users you have, and whether some insist on using old versions of MKCL?
I do not have any counter on the number of MKCL instances in use out there but I had some surprises in the past in that domain. And I am not on notice that anyone insist on using any specific old version of MKCL.
Finally, test-program.script now fails,
I'll look into it (soon I hope). Is this ASDF master head on MKCL master head or otherwise?
I fixed everything in my branch ecl-build. I had to use fas instead of fasb on mkcl. Will investigate at some point. On Fri, Feb 3, 2017, 04:49 Jean-Claude Beaudoin < jean.claude.beaudoin@gmail.com> wrote:
On Thu, Feb 2, 2017 at 10:38 PM, Faré <fahree@gmail.com> wrote:
Dear Jean-Claude,
did MKCL recently change the bundle file type from "fasb" to "fas"? ASDF seems to be failing its test-bundle.script now.
No. Nothing has moved in a long time on that front.
For the record, I never liked the idea of .fasb; but if you make incompatible changes, please synchronize with me and other users.
That .fasb business is a legacy of ECL as you know from the history of asdf-bundle. I will duly inform you if I ever purposely consider some incompatible change in that area of MKCL.
Also, it seems to me that si:mkcl-version and cl:lisp-implementation-version used to be the same (and/or the former didn't exist?), but now latter is more precise whereas the former is used for number .so. Is that correct?
Yes, this was the first post-1.1.10 change to MKCL. This change should only affect git controlled development versions of MKCL. For released versions of MKCL the two functions are still identical. Under git, the preference is given to the new #'si::git-describe-this-mkcl instead, which gives a more exact description of the real version of the specific instance of MKCL. I thought that change was a good feature, maybe I was wrong.
Is it OK if I support only the latest version of MKCL?
I don't have any expectations in that area other than that latest ASDF release should work on latest MKCL master branch head or something close to this. Seeing ASDF be backward compatible with earlier MKCL version is surely a nice thing but it is an ASDF initiative.
Do you have an idea of how many users you have, and whether some insist on using old versions of MKCL?
I do not have any counter on the number of MKCL instances in use out there but I had some surprises in the past in that domain. And I am not on notice that anyone insist on using any specific old version of MKCL.
Finally, test-program.script now fails,
I'll look into it (soon I hope). Is this ASDF master head on MKCL master head or otherwise?
(Removing mkcl-announce from Cc. Sorry about that.) Actually, the issue is that I was calling build-fasl but must now call build-bundle when creating a bundle fasl. I now need to distinguish between :fasl and :fasb in a way that I used not to have to (with code inherited from the old asdf-ecl). Now I wonder how it used to work and what changed there exactly. All should be back to normal on my (now misnamed) ecl-build branch. PS: I find weird that some of the *.fasb in contrib/ are upper-case, while others are lower-case. —♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org Endures the belief in Government "our" friend, no reform is possible. Perishes the superstition, only abolition is conceivable. On Fri, Feb 3, 2017 at 3:57 PM, Jean-Claude Beaudoin <jean.claude.beaudoin@gmail.com> wrote:
On Fri, Feb 3, 2017 at 6:56 AM, Faré <fahree@gmail.com> wrote:
I fixed everything in my branch ecl-build.
I had to use fas instead of fasb on mkcl. Will investigate at some point.
Bizarre, I am clueless as to why this had to be done.
PS: I find weird that some of the *.fasb in contrib/ are upper-case, while others are lower-case.
To fix test-require.script (and the underlying functionality), I had to upcase the modules I'm requiring from mkcl, whereas the downcase version used to work. (Happily, asdf still works in both cases.) —♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org The early bird gets the worm, but the second mouse gets the cheese. — Stephen Pinker
participants (2)
-
Faré -
Jean-Claude Beaudoin