Re: Benchmarks of guile-www against HB's Guile module.

[Alejandro Forero Cuervo]

Hey.

    my comment was directed at variable names only.  the thinking is that
    "globals are bad" (or at least mostly undesirable for their property of
    changing local behavior from a distance), so it's good to make them
    visually distinct as a warning.  lispish programs conventionally use
    *VAR*.

Ahh! :)

Thank you for the clarification.

I agree  that globals are bad  and in this particular  situation I would
rather not  register the  *hb-file* object  and make  it visible  to the
Guile programmer.   I'm still looking for  a way to associate  a pointer
with a procedure and have it passed automatically by Guile.

    if this appeals to you, you might consider restructuring the (guile)
    interface so that programmers can write something like:
    
    (use-module (hb registrar) (ice-9 common-list))
    
    (define (my-proc-1 whatever args) ...)
    (define (my-proc-2 whatever args) ...)
    ...
    
    (hb-register-procs my-proc-1 my-proc-2)

Hmm.

It shouldn't be overly complicated to create this sort of functionality.
One only  needs to add an  interface to HB  so it can be  controled from
within Guile. :)

Perhaps something like:

    (use-module (hb server))

    ; The following register the site's pages:

    (hb-register main proc0)
    (hb-register foo  proc1)
    (hb-register bar  proc2)

    ; The following makes the server wait for requests and execute
    ; them until the process receives SIGTERM.

    (hb-exec)

    ; Now you add clean-up functions.

When hb-exec is called, HB's  control loop is executed.  Some parameters
might be accepted here (such as,  for instance, a proc that is evaluated
once after every request to decide whether the loop must go on or stop).

Or perhaps the primitives for HB's control loop can be made available to
Guile so one can implement the  control loop based on primitives such as
(hb-wait-for-request timeout) or (hb-exec-request req).

I'm not  sure all this would  be worth it. At  the very least, it  is an
interesting idea to make HB entirely controlable from within Guile code.

    making configuration thus evaluative (as opposed to declarative) can
    be a pain  to support, but has benefits for  the programmer (in this
    example,  they can  use  whatever (ice-9  common-list) provides,  in
    their procs).

I don't know  what ice-9 common-list provides.  Why can't  they use that
from within HB's Guile module as it currently stands?

Thanks. :)

Alejo.
http://bachue.com/alejo

--
The mere formulation of a problem is far more essential than its solution.
      -- Albert Einstein.

$0='!/sfldbi!yjoV0msfQ!sfiupob!utvK'x44;print map{("\e[7m \e[0m",chr ord
(chop$0)-1)[$_].("\n")[++$i%77]}split//,unpack'B*',pack'H*',($F='F'x19).
"F0F3E0607879CC1E0F0F339F3FF399C666733333CCF87F99E6133999999E67CFFCCF3".
"219CC1CCC033E7E660198CCE4E66798303873CCE60F3387$F"#Don't you love Perl?

pgp0_xsBXZkNq.pgp
Description: PGP signature