Hi, I updated slime from git today and have been having problems with it (linux x86 and x86-64 sbcl). When I use quicklisp to load drakma it gets somewhere in flexi-streams and then hangs, the sbcl process stops using any cpu and I can't interrupt with C-cC-c. I can use the restart-inferior-lisp command from the repl, and if I do so, then at the second attempt to quickload it gets further, but hangs again further along. I assume that this is because it re-uses some of the fasls from the previous compilation, because if I delete the ~/.cache/common-lisp/ directory then it behaves as before. I attach a sample session, you don't see it, but each time I got a prompt before the very last I had to use restart-inferior-lisp. It's important to state that if I start sbcl from a terminal, then it successfully quickoads drakma, even if I delete the fasl cache, so there must be some interaction that is going wrong between sbcl and slime. --- This email is free from viruses and malware because avast! Antivirus protection is active. http://www.avast.com
On Wed, Feb 19, 2014 at 7:26 PM, Peter Stirling <peter@pjstirling.plus.com> wrote:
I updated slime from git today and have been having problems with it (linux x86 and x86-64 sbcl). When I use quicklisp to load drakma it gets somewhere in flexi-streams and then hangs, the sbcl process stops using any cpu and I can't interrupt with C-cC-c.
FWIW, I can reproduce this. The issue seems to have been introduced by one of the commits pushed today (Feb 19th). -- Luís Oliveira http://kerno.org/~luis/
On Wed, Feb 19, 2014 at 8:16 PM, Luís Oliveira <luismbo@gmail.com> wrote:
FWIW, I can reproduce this. The issue seems to have been introduced by one of the commits pushed today (Feb 19th).
More precisely, git bisect blames this commit: https://github.com/slime/slime/commit/74ef7010d79fad14db43d6ff989f177a7a8986... -- Luís Oliveira http://kerno.org/~luis/
On Wed, Feb 19 2014, Luís Oliveira wrote:
On Wed, Feb 19, 2014 at 7:26 PM, Peter Stirling <peter@pjstirling.plus.com> wrote:
I updated slime from git today and have been having problems with it (linux x86 and x86-64 sbcl). When I use quicklisp to load drakma it gets somewhere in flexi-streams and then hangs, the sbcl process stops using any cpu and I can't interrupt with C-cC-c.
FWIW, I can reproduce this. The issue seems to have been introduced by one of the commits pushed today (Feb 19th).
Presumably caused by deadlocks with SBCL when a function called from the compiler waits for something (as quicklisp does with the macroexpand-hook) and no other thread can call generic functions because the compiler is blocked. I restored the old implementation and it seems to work again. Helmut
On Wed, Feb 19, 2014 at 8:46 PM, Helmut Eller <eller.helmut@gmail.com> wrote:
Presumably caused by deadlocks with SBCL when a function called from the compiler waits for something (as quicklisp does with the macroexpand-hook) and no other thread can call generic functions because the compiler is blocked.
It might have something to do with the compiler lock. Perhaps the first call to one of the generic functions is triggering compilation? -- Luís Oliveira http://kerno.org/~luis/
On Wed, Feb 19, 2014 at 9:48 PM, Luís Oliveira <luismbo@gmail.com> wrote:
On Wed, Feb 19, 2014 at 8:46 PM, Helmut Eller <eller.helmut@gmail.com> wrote:
Presumably caused by deadlocks with SBCL when a function called from the compiler waits for something (as quicklisp does with the macroexpand-hook) and no other thread can call generic functions because the compiler is blocked.
It might have something to do with the compiler lock. Perhaps the first call to one of the generic functions is triggering compilation?
Sorry, I didn't read your message carefully enough. That's precisely what you described. -- Luís Oliveira http://kerno.org/~luis/
participants (3)
-
Helmut Eller -
Luís Oliveira -
Peter Stirling