From nathan at acceleration.net Wed Oct 22 16:49:10 2008 From: nathan at acceleration.net (Nathan Bird) Date: Wed, 22 Oct 2008 12:49:10 -0400 Subject: [cl-containers-devel] On Containers, binary locations, and system connections Message-ID: <48FF5986.6000905@acceleration.net> Most directly: CL-containers defines the following system connection (asdf:defsystem-connection containers-and-utilities :requires (cl-containers metatilities-base) :components ((:module "dev" :components ((:file "utilities-integration"))))) However since the cl-containers system itself depends on metatilities base it appears that this system connection is redundant. and that file should probably just be in the containers system proper. How I came across this: Trying to setup a buildbot running under its own user, it doesn't have access to write fasls to /usr/local/lib/sbcl/site/... AHA I said, let me go grab that binary-locations library. This seems to be working for the most part except that we carve out /usr/local/lib/sbcl as a location that it shouldn't alter. I didn't think that would be a problem since everything in that directory I consider to be already built. System-connections though, when it loads the system, uses ':force t' to force recompilation (unnecessarily? I suspect that there at least was a reason at one point, can it go away?). I tried setting the asdf:*source-to-target-mappings* to nil so that it would compile everything to the private location, which of course lead to the regression that was designed to prevent. I also tried (setf asdf:*source-to-target-mappings* (list (list "/usr/local/lib/sbcl/site" ;;asdf-install locations should be recompilable (concatenate 'string asdf:*default-toplevel-directory* (asdf:implementation-specific-directory-name) "/usr/local/lib/sbcl/site")) '("/usr/local/lib/sbcl" NIL))) ;;sbcl-sourced packages should be left alone. That configuration apparently didn't work--"failed to find the TRUENAME of /home/buildmaster/.sbcl/fasls/sbcl-1.0.21-linux-x86/usr/local/lib/sbcl/sb-posix/constants.lisp-temp: No such file or directory" So it doesn't appear to be properly separating sbcl/ and sbcl/site even though the code does appear to search from front to back. I'm stuck at the moment, will try to dig deeper a bit later. Nathan Bird From attila.lendvai at gmail.com Wed Oct 22 17:15:02 2008 From: attila.lendvai at gmail.com (Attila Lendvai) Date: Wed, 22 Oct 2008 19:15:02 +0200 Subject: [cl-containers-devel] On Containers, binary locations, and system connections In-Reply-To: <48FF5986.6000905@acceleration.net> References: <48FF5986.6000905@acceleration.net> Message-ID: > System-connections though, when it loads the system, uses ':force t' to > force recompilation (unnecessarily? I suspect that there at least was a > reason at one point, can it go away?). iirc, once Gary pushed a patch that removed the :force but there's some other bug (?) somewhere else which steps in and iirc, it results in silently not loading system connections then. it would be nice to find that other bug, because it's also a small annoyance that all system connections are always recompiled. -- attila ps: asdf-system-connections is not in the darcs repo list at cl.net, so i can't easily look up the history for this. From nathan at acceleration.net Wed Oct 22 18:24:08 2008 From: nathan at acceleration.net (Nathan Bird) Date: Wed, 22 Oct 2008 14:24:08 -0400 Subject: [cl-containers-devel] On Containers, binary locations, and system connections In-Reply-To: References: <48FF5986.6000905@acceleration.net> Message-ID: <48FF6FC8.7020406@acceleration.net> An HTML attachment was scrubbed... URL: From attila.lendvai at gmail.com Thu Oct 23 00:42:33 2008 From: attila.lendvai at gmail.com (Attila Lendvai) Date: Thu, 23 Oct 2008 02:42:33 +0200 Subject: [cl-containers-devel] On Containers, binary locations, and system connections In-Reply-To: <48FF6FC8.7020406@acceleration.net> References: <48FF5986.6000905@acceleration.net> <48FF6FC8.7020406@acceleration.net> Message-ID: > But that was the only reference I or my google-fu could produce. I don't see > anything on c-l.net mailing lists from 12/2006 or 1/2007, I tried looking at > cl-containers, asdf-*, metatilities-devel, moptilities-devel. here's my original mail regarding this. unfortunately i don't remember any more details than available here: ---------- Forwarded message ---------- From: Attila Lendvai Date: Fri, Jan 19, 2007 at 5:18 PM Subject: Re: some more bind patches available To: Gary King > i've got some fixes available for bind at: i've pushed two more patches. and unfortunately there's also an issue: asdf-system-connection seems to be broken. an example is bind whose system-connections defined in its .asd are not executed. unpulling the patch "Removed an errant :force keyword in the call to asdf:oos" fixed it for me. maybe i'm missing something in my version of asdf? (sbcl cvs) unpulling this brought back the effect that the contextl integration of metacopy is never loaded from fasl's, but always fresh-compiled. but at least it's loaded now. hth, -- - attila "- The truth is that I've been too considerate, and so became unintentionally cruel... - I understand. - No, you don't understand! We don't speak the same language!" (Ingmar Bergman - Smultronst?llet) From gwking at metabang.com Thu Oct 23 02:34:25 2008 From: gwking at metabang.com (Gary King) Date: Wed, 22 Oct 2008 22:34:25 -0400 Subject: [cl-containers-devel] On Containers, binary locations, and system connections In-Reply-To: References: <48FF5986.6000905@acceleration.net> <48FF6FC8.7020406@acceleration.net> Message-ID: Hi Nathan and Attila, Thanks for bringing this up; I'll try investigating it in some detail tomorrow. On Oct 22, 2008, at 8:42 PM, Attila Lendvai wrote: >> But that was the only reference I or my google-fu could produce. I >> don't see >> anything on c-l.net mailing lists from 12/2006 or 1/2007, I tried >> looking at >> cl-containers, asdf-*, metatilities-devel, moptilities-devel. > > > > here's my original mail regarding this. unfortunately i don't remember > any more details than available here: > > > > ---------- Forwarded message ---------- > From: Attila Lendvai > Date: Fri, Jan 19, 2007 at 5:18 PM > Subject: Re: some more bind patches available > To: Gary King > > >> i've got some fixes available for bind at: > > i've pushed two more patches. > > and unfortunately there's also an issue: asdf-system-connection seems > to be broken. an example is bind whose system-connections defined in > its .asd are not executed. unpulling the patch "Removed an errant > :force keyword in the call to asdf:oos" fixed it for me. maybe i'm > missing something in my version of asdf? (sbcl cvs) > > unpulling this brought back the effect that the contextl integration > of metacopy is never loaded from fasl's, but always fresh-compiled. > but at least it's loaded now. > > hth, > > -- > - attila > > "- The truth is that I've been too considerate, and so became > unintentionally cruel... > - I understand. > - No, you don't understand! We don't speak the same language!" > (Ingmar Bergman - Smultronst?llet) > _______________________________________________ > cl-containers-devel mailing list > cl-containers-devel at common-lisp.net > http://common-lisp.net/cgi-bin/mailman/listinfo/cl-containers-devel -- Gary Warren King, metabang.com Cell: (413) 559 8738 Fax: (206) 338-4052 gwkkwg on Skype * garethsan on AIM From ssbm2 at o2.pl Tue Oct 28 22:10:46 2008 From: ssbm2 at o2.pl (Szymon) Date: Tue, 28 Oct 2008 23:10:46 +0100 Subject: [cl-containers-devel] newbie problem with search-for-item wihtout :key & :test. Message-ID: <49078DE6.7090409@o2.pl> ;;; Hi, ;; I just asdf-ed cl-containers, and my first example failed ;( ;; Imho according to docs this should work (require :cl-containers) (use-package :metabang.cl-containers) (defclass x () ((letter :accessor x-letter :initarg :letter :type string) (name :accessor x-name :initarg :name :type string))) (defparameter *container-of-x* (make-container 'list-container :key #'x-letter :test #'string=)) (insert-list *container-of-x* (map 'list (lambda (c) (make-instance 'x :letter (string c) :name (char-name c))) "?????????????????????????????????")) ;; ;; OK: ;; (search-for-item *container-of-x* "?" :test #'string= :key #'x-letter) ;; ;; FAILs: ;; (search-for-item *container-of-x* "?") ;; The value # ;; is not of type ;; (OR (VECTOR CHARACTER) (VECTOR NIL) BASE-STRING SYMBOL ;; CHARACTER). ;; [Condition of type TYPE-ERROR] ;; ;; Restarts: ;; 0: [ABORT] Return to SLIME's top level. ;; 1: [TERMINATE-THREAD] Terminate this thread (#) ;; ;; Backtrace: ;; 0: (STRING= "?" #)[:EXTERNAL] ;; 1: (FIND "?" (# # # # # # ...))[:EXTERNAL] ;; Locals: ;; SB-DEBUG::ARG-0 = 6 ;; SB-DEBUG::ARG-1 = "?" ;; SB-DEBUG::ARG-2 = (# # # # # # ...) ;; Regards, Szymon. From gwking at metabang.com Wed Oct 29 02:47:41 2008 From: gwking at metabang.com (Gary King) Date: Tue, 28 Oct 2008 22:47:41 -0400 Subject: [cl-containers-devel] newbie problem with search-for-item wihtout :key & :test. In-Reply-To: <49078DE6.7090409@o2.pl> References: <49078DE6.7090409@o2.pl> Message-ID: <602BBA46-1ABF-4C3E-AE29-280D17838EF5@metabang.com> Hi Thanks for the question, It certainly looks like something has gone awry somewhere. I'll try to look at this in the morning. regards, On Oct 28, 2008, at 6:10 PM, Szymon wrote: > ;;; Hi, > ;; I just asdf-ed cl-containers, and my first example failed ;( > ;; Imho according to docs this should work > > (require :cl-containers) > > (use-package :metabang.cl-containers) > > (defclass x () > ((letter :accessor x-letter :initarg :letter :type string) > (name :accessor x-name :initarg :name :type string))) > > (defparameter *container-of-x* > (make-container 'list-container :key #'x-letter :test #'string=)) > > (insert-list *container-of-x* > (map 'list > (lambda (c) (make-instance 'x :letter (string > c) :name (char-name c))) > > "?????????????????????????????????")) > ;; > ;; OK: > ;; > > (search-for-item *container-of-x* "?" :test #'string= :key #'x- > letter) > > ;; > ;; FAILs: > ;; > > (search-for-item *container-of-x* "?") > > ;; The value # > ;; is not of type > ;; (OR (VECTOR CHARACTER) (VECTOR NIL) BASE-STRING SYMBOL > ;; CHARACTER). > ;; [Condition of type TYPE-ERROR] > ;; > ;; Restarts: > ;; 0: [ABORT] Return to SLIME's top level. > ;; 1: [TERMINATE-THREAD] Terminate this thread (# RUNNING {B59BB21}>) > ;; > ;; Backtrace: > ;; 0: (STRING= "?" #)[:EXTERNAL] > ;; 1: (FIND "?" (# # # # {BC9CDF9}> # # ...))[:EXTERNAL] > ;; Locals: > ;; SB-DEBUG::ARG-0 = 6 > ;; SB-DEBUG::ARG-1 = "?" > ;; SB-DEBUG::ARG-2 = (# # # {BC9CDE9}> # # # ...) > > ;; Regards, Szymon. > > _______________________________________________ > cl-containers-devel mailing list > cl-containers-devel at common-lisp.net > http://common-lisp.net/cgi-bin/mailman/listinfo/cl-containers-devel -- Gary Warren King, metabang.com Cell: (413) 559 8738 Fax: (206) 338-4052 gwkkwg on Skype * garethsan on AIM