May 2015 - asdf-devel - mailman3.common-lisp.net

Declaring code obsolete
by Faré 23 May '15

23 May '15

Dear lispers, I'm looking for hackers to review my branch obsolete-function-warnings: https://gitlab.common-lisp.net/asdf/asdf/tree/obsolete-function-warnings In it, I add support to UIOP for issuing conditions for obsolete functions, and use it for UIOP and ASDF deprecated functions. They can be style-warning, warning, continuable error or error, if possible at compile-time via a compiler-macro, or at first use at runtime. Last option is plain error at compile-time for having failed to delete the functions when promised. There is also a cool but somewhat dangerous optional feature that can automatically bumps the status of an obsolete function at every release — you better not use it lightly. Finally, this was the opportunity to realize that there are plenty of offenders in Quicklisp that use long obsolete APIs of ASDF and need be updated. Yikes. QUESTIONS: * Does it make sense to go from :style-warning to :warning to :error to :delete ? * Does it make sense for :error to be a continuable error, or should I introduce an extra step for :cerror, then an :error that isn't continuable? * Robert is in charge of the release schedule, so he should decide what automatic bumping to use, if any. I suppose he may want style-warning now while we're in 3.1, then warning at 3.2, then error at 3.3, then function deleted in 3.4; except for some recently deprecated functions that should only start issuing style-warning in 3.2, and for some backward compatibility package munging that I hope can go away in 3.3. * At some point soon, someone must contact all those people using obsolete functions and classes, and chide them or offer patches. * Now is a good time to upgrade, since all maintained implementations at long last provide asdf3 (May 2013, so a 2 year delay from release to ubiquity; let's hope 3.1 will be ubiquitous next year). —♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org "To speak algebraically, Mr. M. is execrable, but Mr. G. is (x+1)ecrable." — Edgar Alan Poe

1 0

New version of ASDF pushed
by Robert P. Goldman 19 May '15

19 May '15

I have just pushed 3.1.4.5, which contains fixes to UIOP's configuration file finding/XDG functions needed to fix launchpad bug 1437480. Testing would be very much appreciated, especially by those using the configuration file finding functions.

2 10

Patch to UIOP/LISP-BUILD:GET-OPTIMIZATION-SETTINGS for SBCL
by Douglas Katzman 18 May '15

18 May '15

Hi ASDF developers, I'm changing SB-C::*POLICY* from a list to a struct. (This yields us a nice compile-file speedup.) Instead of accessing the global policy with (CDR (ASSOC ...)) it should be accessed with (SB-C::POLICY-QUALITY) as per the attached diff. This function has been present since 2000-12-29 and the interface is unchanged, so a new ASDF can work just fine with an older SBCL. The reverse is not possible. I'll patch the ASDF that ships with SBCL. Doug

2 2

EXPECTED test failures & clisp
by Robert P. Goldman 07 May '15

07 May '15

Do we have a way of indicating that we expect a test to fail? There's a string in the output that talks about "Unexpected test failures in..." but I am not sure if there's actually any notion of expected vs. unexpected test failure. IIRC in the past when I knew a test would fail I just used reader macros to make sure it wouldn't run. This is actually not as good as having the test run but not cause a failure. I ask because I get test failures in clisp on run-program, because somehow clisp lets the common-lisp prompt leak into the output of running a program. So I get something like ("[4]> hello, world") instead of ("hello, world") IIRC this is a known clisp problem, and may even be fixed in the clisp source. But there hasn't been a clisp release for almost five years now, and I don't intend to build it from source. With no releases since 2010, IMO clisp is only "living dead," and possibly simply "dead."

3 2

Ding dong the witch is dead!
by Faré 06 May '15

06 May '15

Dear lispers, with the recent release of LispWorks 7 (congratulations to the LispWorks team!), each and every maintained CL implementation now provides ASDF 3, and ASDF 2 can be declared officially dead. Adieu, ASDF 2, you served your purpose, but you won't be missed. I propose that free software libraries (including ASDF itself) may safely assume ASDF 3 everywhere and drop support for ASDF 2 or earlier. Note: I didn't count GCL as an actively maintained CL implementation, which is only slightly harsh. Maintained (and ship with asdf 3): abcl ccl clasp clisp cmucl ecl mkcl sbcl allegro lispworks Unmaintained (but have worked at some point with asdf): gcl mcl xcl cormanlisp genera mocl scl https://youtu.be/PHQLQ1Rc_Js —♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org "Computer programming is omnipotence without omniscience." — Prospero, as cited by Eliezer Yudkowsky

2 1

UIOP documentation?
by Jared C. Davis 06 May '15

06 May '15

Hi, The ASDF web site at https://common-lisp.net/project/asdf/ has a very nice manual. Is there similar documentation for UIOP itself? If so, it would be great to add a link to it from the ASDF web page. Cheers, Jared -- Jared C. Davis <jared(a)cs.utexas.edu> 11410 Windermere Meadows Austin, TX 78759 http://www.cs.utexas.edu/users/jared/

4 8

Re: UIOP documentation?
by Robert P. Goldman 06 May '15

06 May '15

Faré wrote: >> > I looked at the link to the version that's supposed to have good docs: >> > >> > http://bimib.disco.unimib.it/people/Marco.Antoniotti/Projects/CL/HELAMBDAP/… >> > >> > AFAICT it shows only the docs for BACKWARD-INTERFACE, and maybe not even >> > all of those. >> > > Oh! It has more documentation than that, but depending on your screen > resolution, it can be totally unobvious that you can scroll the > package pane. Le sigh. I tried to document that in the uiop/README.md, > but that needs also be told in the index.html. > That is a *horrible* left pane. As you say, there is no visible affordance, indicating an ability to scroll that pane. The pane also has no border, which hides the fact that, instead of filling the frame with package entries, it takes only 25% of the height of the pane. I guess that's because it wants to put the package interface below the package menu. A border around those frames would help the user understand how the layout is to be used. Also, one I have selected one package, clicking on another in the sidebar takes a *very* long time (so long that I thought it was not going to update at all) to update the main frame.

1 0

Fwd: UIOP documentation?
by Faré 06 May '15

06 May '15

[I got notification on the below message being rejected because the quickdocs url is blacklisted by http://lookup.uribl.com/ !? I stripped said url so the message makes it to the list.] On Tue, May 5, 2015 at 8:35 PM, Robert P. Goldman <rpgoldman(a)sift.net> wrote: > Faré wrote: >> UIOP has no manual, but it has a README.md which guides you through >> the code and links to a website with interactive help as extracted >> from docstrings. I included a direct link on the ASDF webpage in >> doc/index.html — but only the maintainer has the rights to update the >> actual webpage. > > I looked at the link to the version that's supposed to have good docs: > > http://bimib.disco.unimib.it/people/Marco.Antoniotti/Projects/CL/HELAMBDAP/… > > AFAICT it shows only the docs for BACKWARD-INTERFACE, and maybe not even > all of those. > Oh! It has more documentation than that, but depending on your screen resolution, it can be totally unobvious that you can scroll the package pane. Le sigh. I tried to document that in the uiop/README.md, but that needs also be told in the index.html. > As you report, the quickdocs page, http://....org/uiop/api, only > provides docs for UIOP/PACKAGE. > > Also, for some reason these aren't clickable links in the README.md: > they come through as preformatted text, instead. > OK, I was assuming ReST-style magic. I pushed a fix to that. Actually, I made a few passes over both README.md and uiop/README.md. I also merged into the minimakefile branch. —♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org An anarchist is a man who is careful to always use pedestrian crossings, because he utterly detests talking with policemen. — Georges Brassens

1 0

Lisp scripting
by Faré 06 May '15

06 May '15

>> * minimakefile. Can we get this in before we cut 3.2? Actually, >> merging before 3.1.5 would allow to test the release functions before >> we release an actual 3.2. > > The last time I was working with minimakefile it was still at the state > where every time I tried a new command, I would get a new error. I got > tired of this and gave up. I'll try again, but this one is your cause: > I don't care if CL never becomes a scripting language. I'm willing to > run with it if it works, but if it gives me grief, I'll give up again. > Indeed, it's my pet topic; I apologize for the bugs, and I am grateful to you for bearing with me thus far. Obviously, there was a big gap from "(seems to) works for me" to "works robustly for everyone"; but this time, I believe the bugs have been shaken, and the new test system is superior in many ways, including not drowning the screen in so much noise, providing better replay information on error, supporting one more Lisp (clasp), an interactive list of commands, ability to invoke tests from the SLIME REPL, etc. Please try again, and merge into master if satisfactory. —♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org When one realises one is asleep, at that moment one is already half-awake. ― Pyotr Uspensky

1 0