[slime-devel] Long running functions

[Elliott Slaughter]

On Mon, Dec 7, 2009 at 2:39 AM, Helmut Eller <heller at common-lisp.net> wrote:

> * Elliott Slaughter [2009-12-07 10:30+0100] writes:
> >
> > I've been reading the manual page on swank communication style, and it
> sounds
> > like the :spawn option does what I want. Unfortunately, thread support is
> not
> > universal, and happens to missing on one of my main development platforms
> > (SBCL on Windows). Is there any way I can work around this (e.g. add
> swank
> > callbacks to my application)? I know this is kind of a long shot but it
> would
> > be kind of nice to get this working independent of implementation thread
> > support.
>
> There's also :sigio which doesn't need threads but probably also not
> available on Windows.
>
> Your application could call swank::handle-requests from time to time.
> The timeout argument should be 0 to poll the socket without blocking.
> :fd-handler style does just that whenever SBCL performs a blocking read.
> For this, read-char-no-hang should work for sockets which was broken the
> last time I checked. Maybe it's fixed now.
>

When I try 0 as timeout, I get this error:

The assertion (OR (NOT SWANK::TIMEOUT) (EQ SWANK::TIMEOUT T)) failed.

When I use t as timeout, it does basically what I want, but I'd rather have
the timeout be immediate (no wait). Is there any way to do this?

Also, it isn't entirely clear what I should pass as a connection to
handle-requests. Right now I'm using (or swank::*emacs-connection*
(swank::default-connection)) which seems to work fine.

Thanks for the help.

>From Emacs, entering the debgger with C-c C-b should work most of the
> time (as usual, may be broken on Windows).  And it should be possible to
> use most commands on top of the debugger.
>
> I think ILisp had a option to always send an interrupt before a command.
> Not sure how well that would work.
>
> There's also CCL which has a less handicapped Windows port.
>


-- 
Elliott Slaughter

"Don't worry about what anybody else is going to do. The best way to predict
the future is to invent it." - Alan Kay
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.common-lisp.net/pipermail/slime-devel/attachments/20091207/d4825a60/attachment.html>