Re: Translation of http status code to text

[Ted Zlatanov]

On Tue, 23 Mar 2010 11:54:18 +0200 Juri Linkov <address@hidden> wrote: 

JL> Please don't add symbols.  Developers prefer to refer to numbers.

I considered that but it really seemed useful to have symbolic codes as
I was working through the list of text explanations.

On Tue, 23 Mar 2010 11:54:40 +0100 address@hidden wrote: 

j> As a developer I'd like both numbers and symbols. Maybe thats just me. I
j> think this is a bit particular for the http protocol, since some codes
j> are extremely well known, whereas others are not.

Ditto.

On Tue, 23 Mar 2010 08:57:15 -0400 Stefan Monnier <address@hidden> wrote: 

SM> I see, then it's OK to add symbols, I guess, but in this case use
SM> shorter ones, so you can do:

SM>   (case (cadr (assq status-number url-http-codes)))
SM>     (OK ...)
SM>     (moved-permanently ...)
SM>     (proxy-authentication-required ...)
SM>     (accepted ...)
SM>     ...)

Yeah, that was the other option.  It works too.

SM> And in that case, a better option is to create the symbols
SM> programmatically from the error string:

SM>   (defconst url-http-codes
SM>     (mapcar (lambda (x)
SM>               (let ((s (subst-char-in-string ?\s ?- (cadr x))))
SM>                 (when (string-match "-(.*)" s)
SM>                   (setq s (substring s 0 (match-beginning 0))))
SM>                 (list (car x) (intern (downcase s)) (cadr x))))
SM>      '((100 "Continue with request")
SM>        ...)))

>From the perspective of the programmer it's better to see the symbols
(grep won't work with this approach).  Also "OK" look weird as the "ok"
status code and "time-out" in the error code is odd IMHO.  These are
matters of taste but I hope you'll allow that the manually generated
table is a little better.  It only has to be done once, after all.

Attached is another attempt which uses the symbols in the code and
doesn't dynamically generate the symbols.

Ted

url-http-codes.patch
Description: Text Data