[Ecls-list] [creelman.david at gmail.com: ECL serving http via hunchentoot, how do I get it to stay up?]

[David Creelman]

On Fri, Aug 17, 2012 at 01:12:06PM +0200, Juan Jose Garcia-Ripoll wrote:
>    On Thu, Aug 16, 2012 at 11:54 PM, David Creelman
>    <[1]creelman.david at gmail.com> wrote:
> 
>      I've setup a tiny server with ECL and I've quicklisp'ed the hunchentoot
>      package and it works well when I start it from the REPL.
>      However, If I try and run ECL as a background task which in turn is
>      setting up
>      my hunchentoot instance, ECL will stop after a variable amount of time
>      (usually
>      5-10 minutes).
> 
>    It does not break when in foreground? Hmm, do you have some simple steps
>    so that I can reproduce it? Best submitted
>    here*[2]http://sourceforge.net/p/ecls/bugs/ but if it is too troublesome
>    for you, just post them in the mailing list.

Hi Juanjo,
Sorry, it's late, so I'll explain my setup here in a reply.
I have a script called go.lisp which has
(ql:quickload "hunchentoot")
(hunchentoot:start (make-instance 'hunchentoot:easy-acceptor :port 8088)

I run this by doing 
ecl -load go.lisp >& /dev/null

It will go for a while and then stop working.

>From the replies above I've adopted the tmux way of doing things and I now 
detach the session under tmux leave it running and it's been working fine
since I started it at lunch time (now midnight here).

I think by default hunchentoot echoes back info about who is connecting to the
server in the REPL and it must need some input and hence fails somewhere along
the line.

I've not investigated the depths of what is going on. Suprisingly it does seem
to serve up for a little while after ECL/hunchentoot pushes out a log message.
Perhaps the serving thread lasts a little while after ECL stops?

I'm quite happy with the tmux way of doing things. I will try to attach slime
at some point to, that could be quite powerful.

Thanks for getting ECL to the point it's at. It seems to be working quite well
now and is a quite good option on a small memory machine (256Mb). SBCL won't
work because it needs a whole lot of memory and CLISP seems to have several
problems with libs.

>    Juanjo
>    --
>    Instituto de F*sica Fundamental, CSIC
>    c/ Serrano, 113b, Madrid 28006 (Spain)*
>    [3]http://juanjose.garciaripoll.googlepages.com
> 
> References
> 
>    Visible links
>    1. mailto:creelman.david at gmail.com
>    2. http://sourceforge.net/p/ecls/bugs/
>    3. http://juanjose.garciaripoll.googlepages.com/