Maybe starting the server from the REPL is the better idea. Right now I do it through raco (redirecting commands of form `raco pollen ...`, including `raco pollen start`, to a special pollen/raco module that handles them)
On Aug 27, 2014, at 3:12 PM, Jay McCarthy <[email protected]> wrote: > Yes, although I actually assumed that you would start your server from > a real Racket REPL, rather than implement your own REPL, so you > wouldn't need to wait on anything. But what you wrote will work too. > > Jay > > On Wed, Aug 27, 2014 at 4:33 PM, Matthew Butterick <[email protected]> wrote: >> Helpful, thanks. I have my `serve/servlet` within a `start-server` function >> rather than at the top level. Quick experimentation suggests that >> `thread-wait` is needed in this case (?): >> >> (define (start-server) >> (setup-chores ...) >> (define server-t (thread (lambda () (serve/servlet ...)))) >> (define repl-t (thread (lambda () ...))) >> (thread-wait server-t)) >> >> Otherwise, I'll see how far I get. >> >> >> >> On Aug 27, 2014, at 10:58 AM, Jay McCarthy <[email protected]> wrote: >> >>> Hi Matthew, >>> >>> You can run serve/servlet in a separate thread and continue to >>> interact with any functions you want: >>> >>> (define server-t (thread (lambda () (serve/servlet ...)))) >>> >>> You could then use something like xrepl to start doing new stuff. >>> There's nothing the Web server would need to do necessarily. >>> >>> However, if you wanted to change how the server was running, you might >>> need to adjust some things. For instance, you could put the main >>> request->response function in a box that you could change. You could >>> kill the server-t thread and then create another one to restart the >>> server with new parameters. >>> >>> Is that a useful sketch? >>> >>> Jay >>> >>> >>> On Wed, Aug 27, 2014 at 1:33 PM, Matthew Butterick <[email protected]> wrote: >>>> The main engine of my Pollen system is a project webserver that's started >>>> from the command line, uses serve/servlet, and sends the usual status & >>>> error messages to the terminal window. >>>> >>>> I was thinking it would be handy to be able to keep an input prompt in >>>> that terminal window to be able to issue commands without stopping the >>>> server, switching to another window, etc. >>>> >>>> Is there a straightforward approach to this in terms of program design, or >>>> would it introduce annoying complications? >>>> >>>> My understanding of serve/servlet is that it can only do one thing, which >>>> is listen & respond to TCP requests on a certain port. So for this >>>> REPL-ish feature, there would have to be a separate listenener/response >>>> mechanism. Is that something one would accomplish with threads? Or is >>>> there a better way? (Or no way.) >>>> >>>> An approach I've seen with other web-development systems is to put an >>>> admin panel in the browser (by injecting it into the HTML of the output >>>> web page). But that's inherently hinky because it pollutes your output. >>>> ____________________ >>>> Racket Users list: >>>> http://lists.racket-lang.org/users >>> >>> >>> >>> -- >>> Jay McCarthy >>> http://jeapostrophe.github.io >>> >>> "Wherefore, be not weary in well-doing, >>> for ye are laying the foundation of a great work. >>> And out of small things proceedeth that which is great." >>> - D&C 64:33 >> >> >> ____________________ >> Racket Users list: >> http://lists.racket-lang.org/users > > > > -- > Jay McCarthy > http://jeapostrophe.github.io > > "Wherefore, be not weary in well-doing, > for ye are laying the foundation of a great work. > And out of small things proceedeth that which is great." > - D&C 64:33 ____________________ Racket Users list: http://lists.racket-lang.org/users
