libera/#commonlisp - IRC Chatlog
Search
8:17:23
specbot
Sections Not Formally Part Of This Standard: http://www.lispworks.com/reference/HyperSpec/Body/01_dc.htm
9:55:42
madnificent
I'm catching errors when constructing an answer for a web service. However, I'd like the full error stack trace to be printed on STDOUT in that case for development purposes. If I hook slime onto the image then I can get a full stack trace, but how do I print that out when catching the error.
9:56:28
madnificent
I'm on SBCL normally and it's fine if it only works there. I've tried using trivial-backtrace but that prints substantially less depth than what I get through slime.
10:05:34
jackdaniel
you are probably doing something like (handler-case (handle) (error () (trivial-backtraceā¦))
10:06:02
jackdaniel
what you want is handler-bind that does not unwind the stack before passing the control to the handler
12:24:39
dbotton
If asking why I may want, testing for reproducible builds probably most common reason for me
13:28:20
mrcom
dbotton: are you dropping into ldb? Ouch. That sounds like a heap-exhaustion or we-did-something-not-thread-safe-and-now-we're-toast.
13:32:01
mrcom
Just looked at the HT docs, and it says the default Taskmaster creates a thread for each connection. So killing a thread won't run out of task handlers.
18:46:57
pjb
alethkit: but even the implementations targetting POSIX systems, they mostly run on bare hardware, with just a few dependencies in libc and a kernelā¦ They could easily be made to run on raw bare hardware.
19:05:45
rotateq
alethkit: SBCL is an implementation, ran from the start on bare metal afaik and compiles natively.
19:07:09
Bike
"bare metal" can also mean independently of an operating system, and sbcl can't do that
19:13:24
Nilby
I don't think it's even close to easy to run SBCL without an OS. It's tough to get it to be usable with just a kernel. Mezzano is far ahead in that regard.
19:18:22
alethkit
So getting SBCL to work in Mezzano, or getting SBCL to work for embedded development via Mezzano's compiler?
19:40:34
mrcom
Just saw Eitaro Fukamachi's web post (dated today), and he's saying that Hunchentoot is creating a thread per request. HT docs I just saw say, by default, it's per connection. Tom-A-to/Tom-AH-to, outdated info, or ???.
19:43:01
mrcom
n/m--Tom-A-to/Tom-AH-to. Article later goes into detail, says "...creates a new thread when accepting a connection,... terminates the thread when it's disconnected."