From: Daniel Barlow
Subject: SBCL native threads paper; low-key Lisp evangelism at other people's conferences
Date:
Message-ID: <87ptjk88oq.fsf@noetbook.telent.net>
Last week I presented a paper on the threading implementation in SBCL
at the UK Unix Users Group Linux 2003 conference.
<http://www.ukuug.org/events/linux2003/>
As befits a talk for a Linux developers conference, there's more
implementation detail in it than Lisp code, but comp.lang.lisp readers
may be interested anyway. If nothing else, you can play with the toy
slide generation program (requires CLX) that I wrote to present the
slides
http://www.linux.org.uk/~dan/linux2003/
It may have helped convince a few people that Lisp hackers aren't all
handwavey "value of everything, cost of nothing" types. On the other
hand, I'm sure I managed to confirm a few impressions about tendencies
towards NIH.
I wore my ILC2002 t-shirt on another day of the conference, just to
fly the Lisp flag a little. Got a few questions along the "so what is
it good for" lines: I'm still looking for a concise answer that
doesn't sound patronising ("hard problems" => "anything you manage to
do in Perl/C/C++/Java must therefore be so easy as to be beneath me")
but I tend to talk about server-based applications. The state of the
practice for web programming on Linux is still very much "tail the
error log and insert print statements" or at best "look at the
backtrace in the browser window", so there's a certain appeal to being
able to debug interactively there.
Feedback welcome.
-dan
--
http://www.cliki.net/ - Link farm for free CL-on-Unix resources
From: Andreas Fuchs
Subject: Re: SBCL native threads paper; low-key Lisp evangelism at other people's conferences
Date:
Message-ID: <868yq79446.fsf@boinkine.defun.at>
Today, Daniel Barlow <···@telent.net> wrote:
> It may have helped convince a few people that Lisp hackers aren't all
> handwavey "value of everything, cost of nothing" types. On the other
> hand, I'm sure I managed to confirm a few impressions about tendencies
> towards NIH.
Good news, indeed.
> I wore my ILC2002 t-shirt on another day of the conference, just to
> fly the Lisp flag a little. Got a few questions along the "so what is
> it good for" lines: I'm still looking for a concise answer that
> doesn't sound patronising ("hard problems" => "anything you manage to
> do in Perl/C/C++/Java must therefore be so easy as to be beneath me")
> but I tend to talk about server-based applications. The state of the
> practice for web programming on Linux is still very much "tail the
> error log and insert print statements" or at best "look at the
> backtrace in the browser window", so there's a certain appeal to being
> able to debug interactively there.
That gave me an interesting idea: how about putting up a Lisp booth (or
general VHLL, if not enough lispers are interested) at the next
linux/unix event, say LinuxTag? (LinuxTag because it's the biggest event
in travelling distance from my current position; and because I think I
still have intact contact to some of its organizers)
I'd like to give a hands-on experience to interested people (and I'm too
nervous when it comes to giving a talk). Who knows, we might even
impress some religious Twisted Framework fans with lisp's outstanding
debugging and incremental development ability (-:
> Feedback welcome.
Right.
Unlimited numbers of lispers in the general area of (or with ability to
travel to) Germany/Karlsruhe wanted. Must be able to hack on
request. Speech impediments good as punning material. Social skills a
plus. Should prefer to hack when given the choice between ranting and
hacking. Apply via e-mail or follow-up. Deadline: sometime before June
2004.
Hm. The subject was low-key, while my approach sounds like brute
force. I'm in doubt. Therefore, I choose brute force (-:
--
Andreas Fuchs, <···@acm.org>, ···@jabber.at, antifuchs
From: Daniel Barlow
Subject: Re: SBCL native threads paper; low-key Lisp evangelism at other people's conferences
Date:
Message-ID: <87znin7lrq.fsf@noetbook.telent.net>
Andreas Fuchs <···@void.at> writes:
> travel to) Germany/Karlsruhe wanted. Must be able to hack on
> request. Speech impediments good as punning material. Social skills a
> plus. Should prefer to hack when given the choice between ranting and
> hacking. Apply via e-mail or follow-up. Deadline: sometime before June
> 2004.
I'd like to do LinuxTag anyway sometime, so I'd probably be up for
this.
> Hm. The subject was low-key, while my approach sounds like brute
> force. I'm in doubt. Therefore, I choose brute force (-:
Low key _for me_, still involves standing in front of 60 people asking
"anyone do any SPARC hacking? OK, a few people. What is up witht he
SA_SIGINFO handling ther? _Please_, get them to make it do something
normal..."
I'm sure some people thought I was being terrifically unsubtle.
-dan
--
http://www.cliki.net/ - Link farm for free CL-on-Unix resources
From: Ng Pheng Siong
Subject: Re: SBCL native threads paper; low-key Lisp evangelism at other people's conferences
Date:
Message-ID: <bgpm1a$d3h$1@mawar.singnet.com.sg>
According to Andreas Fuchs <···@void.at>:
> Today, Daniel Barlow <···@telent.net> wrote:
> > It may have helped convince a few people that Lisp hackers aren't all
> > handwavey "value of everything, cost of nothing" types. On the other
> > hand, I'm sure I managed to confirm a few impressions about tendencies
> > towards NIH.
>
> Good news, indeed.
Heh heh. I was trained from young to read the first or second sentence
of a paragraph as the "topic" sentence.
However, reading followups in Usenet, somehow I've acquired the habit that
the topic sentence is the _last_ sentence being quoted, so it sounds like
you're saying "good news" to "perpetuating NIH."
;-)
--
Ng Pheng Siong <····@netmemetic.com>
http://firewall.rulemaker.net -+- Manage Your Firewall Rulebase Changes
http://www.post1.com/home/ngps -+- Open Source Python Crypto & SSL
From: Andreas Fuchs
Subject: Re: SBCL native threads paper; low-key Lisp evangelism at other people's conferences
Date:
Message-ID: <86u18v708q.fsf@boinkine.defun.at>
Today, Ng Pheng Siong <····@netmemetic.com> wrote:
> According to Andreas Fuchs <···@void.at>:
>> Today, Daniel Barlow <···@telent.net> wrote:
>>> It may have helped convince a few people that Lisp hackers aren't
>>> all handwavey "value of everything, cost of nothing" types. On the
>>> other hand, I'm sure I managed to confirm a few impressions about
>>> tendencies towards NIH.
>>
>> Good news, indeed.
>
> Heh heh. I was trained from young to read the first or second sentence
> of a paragraph as the "topic" sentence.
>
> However, reading followups in Usenet, somehow I've acquired the habit
> that the topic sentence is the _last_ sentence being quoted, so it
> sounds like you're saying "good news" to "perpetuating NIH."
Right; I guess that sentence looks awkward because I last-minute edited
out a comment about whose NIH tendencies he was referring to. Err,
well. Um.
So, are you coming to LinuxTag? (-:
--
Andreas Fuchs, <···@acm.org>, ···@jabber.at, antifuchs
From: Lars Brinkhoff
Subject: Re: SBCL native threads paper; low-key Lisp evangelism at other people's conferences
Date:
Message-ID: <85n0en8grv.fsf@junk.nocrew.org>
Daniel Barlow <···@telent.net> writes:
> Last week I presented a paper on the threading implementation in SBCL
> at the UK Unix Users Group Linux 2003 conference.
So how did the audience react? Was the talk well received?
--
Lars Brinkhoff, Services for Unix, Linux, GCC, PDP-10, HTTP
Brinkhoff Consulting http://www.brinkhoff.se/
From: Daniel Barlow
Subject: Re: SBCL native threads paper; low-key Lisp evangelism at other people's conferences
Date:
Message-ID: <87r83z6nii.fsf@noetbook.telent.net>
Lars Brinkhoff <·········@nocrew.org> writes:
> Daniel Barlow <···@telent.net> writes:
>> Last week I presented a paper on the threading implementation in SBCL
>> at the UK Unix Users Group Linux 2003 conference.
>
> So how did the audience react? Was the talk well received?
The paper as submitted had the title
Adding native thread support to SBCL
- or -
$n$ things every programmer should know about signal handling
However, for reasons of space (presumably), the talk was listed in the
programme as "$n$ things every programmer should know about signal
handling". So, there were probably at least a few people there that
didn't realise it was going to be a Lisp talk, and I thought I'd
better point this out upfront. I think a couple of people (of about
60) left at this news, which was fair enough.
A few interesting questions. Someone asked if we planned to implement
an emacs: I said we'd had one (Hemlock) for years already. A short
debate on the wisdom of using SIGSEGV for "things it's not supposed to
do" (as in, anything other than kill the program - SBCL's GC uses it
for a write barrier in old generations). That's a tricky one in that
nowhere is it written down that SIGSEGV is intended to be a
recoverable error - on the other hand, it plainly /is/ a recoverable
error as far as the kernel is concerned, apparently for applications
like this - the only reason that things blow up is that third party
libraries install signal handlers which assume that things have blown
up already.
I wasn't tarred and feathered. I did get a round of appluase. Nobody
stood up. We'll have to wait for the feedback forms to come in to
find out what the overall opinion was, but I'm fairly positive - if
I've convinced a few application/library developers to think carefully
when they create interfaces that require #define, or write library
implementations that steal signals from the application, I'll consider
that a success.
-dan
--
http://www.cliki.net/ - Link farm for free CL-on-Unix resources
From: Paolo Amoroso
Subject: Re: SBCL native threads paper; low-key Lisp evangelism at other people's conferences
Date:
Message-ID: <KykxPxCf=26seKBB4du1JgHkmp3P@4ax.com>
On Wed, 06 Aug 2003 12:20:21 +0100, Daniel Barlow <···@telent.net> wrote:
> The paper as submitted had the title
>
> Adding native thread support to SBCL
> - or -
> $n$ things every programmer should know about signal handling
>
> However, for reasons of space (presumably), the talk was listed in the
> programme as "$n$ things every programmer should know about signal
> handling". So, there were probably at least a few people there that
> didn't realise it was going to be a Lisp talk, and I thought I'd
You could have titled the paper "Sexp", and start the talk by saying "now
that I have your attention...".
> A few interesting questions. Someone asked if we planned to implement
> an emacs: I said we'd had one (Hemlock) for years already. A short
McCLIM's Goatee is another--possibly partial--attempt.
Paolo
--
Paolo Amoroso <·······@mclink.it>
From: Paolo Amoroso
Subject: Re: SBCL native threads paper; low-key Lisp evangelism at other people's conferences
Date:
Message-ID: <ke8vP7Cf1EyQPUk3afRZS3busWRE@4ax.com>
On Tue, 05 Aug 2003 15:45:25 +0100, Daniel Barlow <···@telent.net> wrote:
> fly the Lisp flag a little. Got a few questions along the "so what is
> it good for" lines: I'm still looking for a concise answer that
> doesn't sound patronising ("hard problems" => "anything you manage to
What about "you tell me"?
Paolo
--
Paolo Amoroso <·······@mclink.it>