[asdf-devel] Re: software failures (with asdf 3?)
On Sun, Jul 7, 2013 at 8:12 AM, Zach Beane <xach@xach.com> wrote:
I upgraded my dist construction system to asdf 3.0 and get these failures:
- exscribe: Don't know how to normalize (:FEATURE :EXSCRIBE-TYPESET :CL-TYPESETTING). More info at http://report.quicklisp.org/exscribe/2013-07-07/failtail.txt
Exscribe is working well for me. The :FEATURE feature is an ASDF feature that was simultaneously broken and undocumented in late ASDF1, has been fixed but cerror'ed during the ASDF 2 days, and fixed and tested to work before ASDF3 (test in place since 2.31.2). Your backtrace suggests that the error happens in a package called DEPCHECK, that doesn't know about this feature. I suppose you wrote that package and need to teach it about :FEATURE. That's where I realize that the :FEATURE feature is quite misdocumented in the asdf.texinfo in the grammar production "dependency-def" (as opposed to the very different, documented but deprecated feature in "requirement"). Oh well. Not my problem anymore. Robert, you might want to look into this: * in dependency-def, it should be (:feature @var{feature-expression} @var{dependency-def}) * the required-op production is redundant, since it's better covered by the second branch of requirement. * have a quick look at find-component.lisp to make sure the grammar is properly documented; maybe document the new extensible protocol for naming components.
- cl-launch: Could not load ASDF 2.27. More info at http://report.quicklisp.org/cl-launch/2013-07-07/failtail.txt
Looks like you're using cl-launch 3.21.3 or earlier. Please upgrade to the latest 3.22, available since last May (actually same as 3.21.7 from February). —♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org There is never a better measure of what a person is than what he does when he's absolutely free to choose." — William M. Bulger
Faré wrote:
Robert, you might want to look into this: * in dependency-def, it should be (:feature @var{feature-expression} @var{dependency-def}) * the required-op production is redundant, since it's better covered by the second branch of requirement. * have a quick look at find-component.lisp to make sure the grammar is properly documented; maybe document the new extensible protocol for naming components.
I see the first two issues with the manual. I will apply a fix. The last I will have a look at. I may have to come back to you with some follow-up questions. Zach, is there any way we could somehow provide a better API for your DEPCHECK to use, so that you wouldn't have to worry about it getting out of sync with ASDF? Suggestions welcome. Best, r
Robert Goldman wrote:
Robert, you might want to look into this: * in dependency-def, it should be (:feature @var{feature-expression} @var{dependency-def})
OK, I thought I understood this, but I don't. There's a second argument to :feature when used in a dependency-def, but not when it's used in a dependency? So we would see something like: :depends-on (:cl-ppcre (:feature :clx "clx-dependent-library")) and the semantics is If (member :clx *features*) then clx-dependent-library is a dependency. Correct? Is there some reason why this isn't :if-feature or :when-feature? Using :feature as a conditional seems odd.... thanks, r
On Tue, Jul 9, 2013 at 2:41 PM, Robert Goldman <rpgoldman@sift.info> wrote:
Robert Goldman wrote:
Robert, you might want to look into this: * in dependency-def, it should be (:feature @var{feature-expression} @var{dependency-def})
OK, I thought I understood this, but I don't.
There's a second argument to :feature when used in a dependency-def, but not when it's used in a dependency?
In a :in-order-to spec, asdf:feature (not keyword:feature) can be used instead of an operation specifier, to specify that the plan should fail if the feature isn't present, which is then captured by if-component-fails. This is a deprecated feature, with a braindead design, and only remains as a compatibility mode for old versions of SBCL's sb-bsd-sockets and sb-rotate-byte. I recommend you altogether purge this mis-feature in a year or two.
So we would see something like:
:depends-on (:cl-ppcre (:feature :clx "clx-dependent-library"))
and the semantics is
If (member :clx *features*) then clx-dependent-library is a dependency.
Correct?
Yes.
Is there some reason why this isn't :if-feature or :when-feature? Using :feature as a conditional seems odd....
That's how it was named in e7e468 when it was introduced, though with a buggy implementation. Renaming it :when-feature may make sense. Note that I called the component-if-feature slot keyword :if-feature, which you may want to either distinguish yourself from or copy. You could even add an optional else-clause to an :if-feature. Considering how it never worked until ASDF3, I might be the only one to ever have used it, in exscribe, and I'll happily stop using it while you either transition or purge the feature altogether. It's only really useful at for a system's :depends-on. —♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org The worst thing about totalitarian regimes is not that they make people poor, miserable and unfree — it's that they corrupt people's souls, and turning everyone into a double-thinking, double-speaking liar for the sake of survival.
participants (2)
-
Faré -
Robert Goldman