- slime-devel - mailman3.common-lisp.net

Loading and starting Swank on a lisp image
by Mauro D'Agostino 22 Jun '18

22 Jun '18

I'm trying to add a swank server on all of our lisp apps running on images. I want to have everything compiled and loaded, so when I start up the image I don't need to load any .fasl files. I'm using ASDF to load swank and then running: (swank-loader:init :reload t :load-contribs t :quiet nil) On image start up, I'm running: (setf swank::*loopback-interface* "0.0.0.0") (swank:create-server :port 4005 :dont-close t) Swank starts up fine, but, when the first user tries to connect, it tries to load some .fasl files. ; Fast loading /home/mdagostino/.slime/fasl/2.20/allegro-10.1as-64bit-ics-linux-x86-64/contrib/swank-asdf.fasl ; Loading /home/mdagostino/workspace/Library/world/site/cl-ppcre-1.3.2/cl-ppcre-test.asd ; Loading /home/mdagostino/workspace/Library/world/site/slime-20180308-git/contrib/swank-media.lisp ; Fast loading /home/mdagostino/.slime/fasl/2.20/allegro-10.1as-64bit-ics-linux-x86-64/contrib/swank-trace-dialog.fasl ; Fast loading /home/mdagostino/.slime/fasl/2.20/allegro-10.1as-64bit-ics-linux-x86-64/contrib/swank-package-fu.fasl ; Fast loading /home/mdagostino/.slime/fasl/2.20/allegro-10.1as-64bit-ics-linux-x86-64/contrib/swank-presentations.fasl ; Fast loading /home/mdagostino/.slime/fasl/2.20/allegro-10.1as-64bit-ics-linux-x86-64/contrib/swank-repl.fasl ; Fast loading /home/mdagostino/.slime/fasl/2.20/allegro-10.1as-64bit-ics-linux-x86-64/contrib/swank-macrostep.fasl ; Fast loading /home/mdagostino/.slime/fasl/2.20/allegro-10.1as-64bit-ics-linux-x86-64/contrib/swank-fuzzy.fasl ; Fast loading /home/mdagostino/.slime/fasl/2.20/allegro-10.1as-64bit-ics-linux-x86-64/contrib/swank-util.fasl ; Fast loading /home/mdagostino/.slime/fasl/2.20/allegro-10.1as-64bit-ics-linux-x86-64/contrib/swank-c-p-c.fasl ; Fast loading /home/mdagostino/.slime/fasl/2.20/allegro-10.1as-64bit-ics-linux-x86-64/contrib/swank-fancy-inspector.fasl ; Fast loading /home/mdagostino/.slime/fasl/2.20/allegro-10.1as-64bit-ics-linux-x86-64/contrib/swank-arglists.fasl I tried adding this after doing 'swank:create-server, but it doesn't work: (setf swank::*find-module* #'(lambda (module) (declare (ignore module)) nil)) How can I make it so it doesn't look for those files? I would imagine they have already been compiled and loaded into the image. I'm using 'dumplisp to create the image. Thanks in advance! - Mauro

4 5

Re: License for Slime
by Red Daly 12 Jun '18

12 Jun '18

Oops, meant to reply all. On Tue, Jun 12, 2018, 8:55 AM Red Daly <reddaly(a)gmail.com> wrote: > I got approval to contribute if the new code is in a separate file > released under MIT (and other OSI-approved licenses). Thanks for the > suggestion and willingness to do this. > > Aside from edits to Makefiles, I am not able to modify files that have > been released into the public domain. It also sounds problematic or > impossible to apply a license to code previously released into the public > domain, so my original post to this thread was naive - apologies. > > > On Mon, May 14, 2018, 8:25 PM Red Daly <reddaly(a)gmail.com> wrote: > >> Thanks for the replies. The idea of using a license for new code might >> indeed help. I now have some questions out to the open source team here, >> and I will reply when they get back to me. >> >> For some background reading on public domain software from OSI, I found >> this page informative: https://opensource.org/faq#public-domain >> >> On May 14, 2018 8:23 AM, "Luís Oliveira" <luismbo(a)gmail.com> wrote: >> >> On Mon, May 14, 2018 at 3:23 PM Stelian Ionescu <sionescu(a)cddr.org> >> wrote: >> > Maybe you don't necessarily need to change the licence of existing code. >> > Just like with the legacy code from Spice Lisp and CMUCL which was >> public >> > domain, it should be enough to state that the licence for new code is >> MIT >> and >> > over time the code base would become a mixture, just like SBCL. I guess >> this >> > should be ok to appease lawyers. >> >> I wouldn't have a problem with that. Would it help, Red? >> >> >> -- >> Luís Oliveira >> http://kerno.org/~luis/ >> >> >>

1 0

License for Slime
by Red Daly 15 May '18

15 May '18

Dear Slime maintainers, I would like to contribute some code to the project, but my employer (Google) will only permit you to do so if the project is licensed under an OSI-approved license (https://opensource.org/licenses) I'm not a lawyer, but my understanding is that "public domain" code is more complex legally than code released under one of the OSI licenses. Would the authors consider adding such a license to the project in a LICENSE file? Best, Red

4 6

Pure CL unit tests
by Red Daly 05 May '18

05 May '18

I'm modifying swank-c-p-c.lisp to complete "c:m-v-b" to cl:multiple-value-bind, and I would like to add a few unit tests for the swank-c-p-c code. There are only .el tests as of now, so perhaps there are opinions about including Common Lisp tests in the project. Are there any objections to adding Common Lisp tests? If so, does anyone have preferences for CL tests? (e.g., choice of library, file layout)

4 4

assign lisp source buffers to a specific connection
by Orm Finnendahl 03 Apr '18

03 Apr '18

Hi List, the slime doc specifies that "slime-mode source buffers always use the “default” connection". Since I'm frequently working with multiple common lisp instances it would be handy to be able to assign lisp source buffers to specific connections rather than manually switching the default connection before evaluation. To circumvent this my current workflow uses different emacs instances, but I'd prefer using a single emacs instance with multiple lisp connections. My idea is to implement an interactive command like "slime-assign-connection" when in a source buffer, (or an optional additional global flag specifying to prompt for assigning the source buffer to a lisp connection on eval in case it hasn't yet been assigned). In this case it also would be handy to be able to name each connection individually. I could imagine it'd be not too hard to implement this behaviour and maybe someone can give me a hint to get me started implementing it or warn me about possible pitfalls. -- Orm

1 0

Slime and ASDF
by Attila Lendvai 01 Mar '18

01 Mar '18

[moving this to slime-devel] > I recently banged my head against something similar. Turns out, mcclim loads > swank, so be sure to delete .slime/fasl too. been there... that's why my slime locally uses asdf:apply-output-translations for a few years now. slime used to be proudly ASDF free, but maybe that's something to reconsider, at least with an optional support. this is what i'm running with (which i don't recommend for inclusion in this form): (defun binary-pathname (src-pathname binary-dir) "Return the pathname where SRC-PATHNAME's binary should be compiled." + (declare (ignore binary-dir)) + #+asdf + (asdf:apply-output-translations src-pathname) + #-asdf (let ((cfp (compile-file-pathname src-pathname))) (merge-pathnames (make-pathname :name (pathname-name cfp) :type (pathname-type cfp))))) but if the maintainers are open for this direction, then i may look into packaging up and testing a PR for integrating ASDF properly, and optionally. and not forgetting this either: https://github.com/slime/slime/pull/282/ currently swank can break in funny ways when used together with ASDF. e.g. :depends-on (:swank) can be a source of great bafflement when it reloads swank and undo's half of the contrib's, leaving some stuff crippled. and the .slime/fasl dir can also bite you when you thought you have deleted the fasl's, except one outlier. -- • attila lendvai • PGP: 963F 5D5F 45C7 DFCD 0A39 -- “A teacher is never a giver of truth - he is a guide, a pointer to the truth that each student must find for himself. A good teacher is merely a catalyst.” — Bruce Lee (1940–1973)

4 3

swank sbcl patch
by Douglas Katzman 08 Feb '18

08 Feb '18

Hi, Please apply this diff. There are multiple problems with the existing code. * it seems to believe that DISASSEMBLE-CODE-COMPONENT only works on with non-null %code-debug-info. That's not true; it works on any code object. So depending on your standards, it's excessively wordy to just plain wrong. * (ash (sb-kernel:%code-code-size o) sb-vm:word-shift) causes buffer overrun * sb-disassem::align can not be called See https://bugs.launchpad.net/sbcl/+bug/1748041 Thanks Doug --- a/swank/sbcl.lisp +++ b/swank/sbcl.lisp @@ -1613,18 +1613,7 @@ stack." append (label-value-line i (sb-kernel:code-header-ref o i))) `("Code:" (:newline) , (with-output-to-string (s) - (cond ((sb-kernel:%code-debug-info o) - (sb-disassem:disassemble-code-component o :stream s)) - (t - (sb-disassem:disassemble-memory - (sb-disassem::align - (+ (logandc2 (sb-kernel:get-lisp-obj-address o) - sb-vm:lowtag-mask) - (* sb-vm:code-constants-offset - sb-vm:n-word-bytes)) - (ash 1 sb-vm:n-lowtag-bits)) - (ash (sb-kernel:%code-code-size o) sb-vm:word-shift) - :stream s))))))) + (sb-disassem:disassemble-code-component o :stream s)))))

2 1

[OSX/CMU-CL] Problems with inspector
by Didier Verna 29 Jan '18

29 Jan '18

Hello, while trying to inspect values from Slime in a session with CMUCL 21c, I get the following error. I'm on OSX with CMUCL 21.c from the binary darwin-x86 tarballs. Thanks! Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER:Error (TYPE-ERROR) during printing: #<TYPE-ERROR {49F0548D}> [Condition of type TYPE-ERROR] Restarts: 0: [RETRY] Retry SLIME inspection request. 1: [*ABORT] Return to SLIME's top level. 2: [ABORT] Return to Top-Level. Backtrace: 0: (SWANK/CMUCL::MAILBOX NIL)Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER: NIL is not of type MULTIPROCESSING::PROCESS 1: ((FLET SWANK/BACKEND:SEND) NIL T)Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER: NIL is not of type MULTIPROCESSING::PROCESS 2: ("DEFMETHOD STREAM-WRITE-STRING (SLIME-OUTPUT-STREAM T)")Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER: NIL is not of type MULTIPROCESSING::PROCESS 3: ((METHOD STREAM-WRITE-STRING NIL (SWANK/GRAY::SLIME-OUTPUT-STREAM T)) (#(2 1 3 4 0) . #(#)) #<unused-arg> #<SWANK/GRAY::SLIME-OUTPUT-STREAM {49EFAD3D}> "#<" ...)Type-error in KERNEL::OBJECT-NOT-TYPE-E.. 4: (LISP::WRITE-STRING* "#<" #<SWANK/GRAY::SLIME-OUTPUT-STREAM {49EFAD3D}> 0 2)Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER: NIL is not of type MULTIPROCESSING::PROCESS 5: (LISP::%PRINT-UNREADABLE-OBJECT #<Method-Combination OR (:MOST-SPECIFIC-FIRST) {49ED7BB5}> #<SWANK/GRAY::SLIME-OUTPUT-STREAM {49EFAD3D}> T T ...)Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER: N.. 6: (SWANK::CALL/TRUNCATED-OUTPUT-TO-STRING 200 #<Closure Over Function "DEFUN PREPARE-TITLE" {49EFAB59}> "..")Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER: NIL is not of type MULTIPROCESSING::PRO.. 7: (SWANK::CALL-WITH-BINDINGS ((*PRINT-LINES* . 1) (*PRINT-RIGHT-MARGIN* . 75) (*PRINT-PRETTY* . T) (*PRINT-READABLY*)) #<Closure Over Function "DEFUN PREPARE-TITLE" {49EFAAD9}>)Type-error in KERNEL::OBJ.. 8: (SWANK::ISTATE>ELISP #S(SWANK::ISTATE :OBJECT #<Method-Combination OR # {49ED7BB5}> :VERBOSE NIL :PARTS #() :ACTIONS #() :METADATA-PLIST (:SORT-ORDER # :GROUPING-KIND # :CHECKLIST ...) ...))Type-error.. 9: (SWANK::CALL-WITH-RETRY-RESTART "Retry SLIME inspection request." #<Closure Over Function "DEFSLIMEFUN INIT-INSPECTOR" {49EEAFA9}>)Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER: NIL is not of t.. 10: (SWANK::CALL-WITH-BUFFER-SYNTAX NIL #<Closure Over Function "DEFSLIMEFUN INIT-INSPECTOR" {49EEAF99}>)Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER: NIL is not of type MULTIPROCESSING::PROCESS 11: (SWANK:EVAL-FOR-EMACS (SWANK:INIT-INSPECTOR "*") "COMMON-LISP-USER" 50)Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER: NIL is not of type MULTIPROCESSING::PROCESS 12: (SWANK::PROCESS-REQUESTS T)Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER: NIL is not of type MULTIPROCESSING::PROCESS 13: ("DEFUN HANDLE-REQUESTS")Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER: NIL is not of type MULTIPROCESSING::PROCESS 14: ("DEFUN HANDLE-REQUESTS")Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER: NIL is not of type MULTIPROCESSING::PROCESS 15: ("DEFINTERFACE CALL-WITH-DEBUGGER-HOOK" #<Function SWANK:SWANK-DEBUGGER-HOOK {48107A89}> #<Closure Over Function "DEFUN HANDLE-REQUESTS" {49EE99E1}>)Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER.. 16: (SWANK::CALL-WITH-BINDINGS ((*STANDARD-INPUT* . #) (*STANDARD-OUTPUT* . #) (*TRACE-OUTPUT* . #) (*ERROR-OUTPUT* . #) (*DEBUG-IO* . #<Two-Way Stream, Input = #, Output = #>) ...)Type-error in KERNEL::O.. 17: (SWANK::HANDLE-REQUESTS #<SWANK::SINGLETHREADED-CONNECTION {49D65FFD}> T)Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER: NIL is not of type MULTIPROCESSING::PROCESS 18: (SWANK::INVOKE-OR-QUEUE-INTERRUPT #<Closure Over Function "DEFUN PROCESS-IO-INTERRUPT" {49EE93B1}>)Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER: NIL is not of type MULTIPROCESSING::PROCESS 19: (SWANK::PROCESS-IO-INTERRUPT #<SWANK::SINGLETHREADED-CONNECTION {49D65FFD}>)Type-error in KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER: NIL is not of type MULTIPROCESSING::PROCESS -- Resistance is futile. You will be jazzimilated. Lisp, Jazz, Aïkido: http://www.didierverna.info

3 8

ACL MacOS 10.1 SLIME Inspect bug with SLIME 2.20
by Mark H. David 27 Jan '18

27 Jan '18

The SLIME inspect feature within SLIME in 2.20 does not work with ACL in our testing on version 10.1 on MacOS with "alisp" images, both SMP and non-SMP. The same SLIME inspect feature does work in the previous version, SLIME 2.19. Finally, there's no known problem with SLIME inspect with any other Lisps that we've tried in either of the versions of SLIME (2.19, 2.20), i.e., with CCL and SBCL. To recreate, launch SLIME of the appropriate version (2.20) with the appropriate inferior Lisp (ACL alisp 10.1/MacOS), then run slime inspect on any object. For example, do M-x slime-inspect [ret] () [ret] The above invokes SLIME inspect to inspect the Lisp object nil, entered as (). One could use any Lisp object, e.g., the list '(1 2 3). So, the above command ought to put you in a SLIME inspect buffer to let you inspect the object. Instead, it throws you into a debugger buffer. Then, in addition, it is particularly onerous because the input prompted for by the debugger is not apparently responded to. For example, typing q to quit or 3 to abort does nothing. You apparently can do nothing but manually kill the emacs buffers involved with Lisp/SLIME. See an attached screen shot for a picture of how the debugger buffer looks. The screen shot shows that the debugger is somewhere in no-applicable-method MULTIPROCESSING:PROCESS-PROPERTY-LIST.

2 2

best way to report slime bug reports/issues
by Mark H. David 26 Jan '18

26 Jan '18

Is this list the best way to report bugs and issues with SLIME for everyday users? Is the github repo a better place? Is there some other list? Thanks, Mark

2 1