httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ben Laurie <...@gonzo.ben.algroup.co.uk>
Subject Re: cvs commit: apache/src conf.h http_config.c http_config.h http_core.c http_core.h (fwd)
Date Sat, 27 Jul 1996 17:49:19 GMT
Robert S. Thau wrote:
[Moved from the CVS list ... for those who haven't seen it, Robert is objecting
to my addition of a pointer to the cmd_info structure from cmd_parms]

>   Unfortunately it is also used in set_rlimit. I'm happy to discuss
>   it, though.  It seemed like a useful and harmless addition to me,
>   and it is back compatible, unlike many other changes that I'd really
>   like to make.
> 
> It strikes me as creeping featurism.  The problem is that there was
> *already* a way to do this, via the cmd_info machinery.  Adding
> another, by itself, isn't *so* bad, but once these "harmless
> additions" start to pile up, keeping track of the interactions of all
> the redundant, but slightly different mechanisms, gets to be really
> awkward.  Better not to create the situation in the first place, if
> there's an easy way to avoid it --- and in this case there is.
> 
> Using the preexisting cmd_info instead of your new thing wouldn't add
> a single line of code to your command handlers, or to the command
> tables that invoke them; given that, it's going to be awfully hard to
> convince me that cmd->cmd makes them easier to write.

But that ain't true. cmd_info is already used to hold the offset of the limit
structure. However, it isn't used, so I guess I could replace it with the name
of the directive. Strikes me as naff, though.

> 
>   How about TAKE1|TAKE2? Assuming there are considerably less than 32
>   possibilities, that is.
> 
> This opens a larger can of worms than it appears to --- that idea
> looks fine in this particular case, but combinations like FLAG|TAKE1
> ("On, Off, or maybe something else"?), FLAG|TAKE2 ("A flag and an
> optional argument?  On, Off, or maybe something else, and an optional
> argument"?) or even worse, something like RAW_ARGS|ITERATE, are a lot
> harder to interpret.  Best to leave the command table entry indicating
> a *single* grammar for the arguments --- TAKE1 vs. TAKE2 is really
> just about the only case where the command parser can distinguish the
> alternatives without additional information.

True enough.

> 
> BTW, I notice this is going on on apache-cvs.  I thought the reply-to
> address on CVS update notices was supposed to bounce comments on
> them back to the main list?

Speak to Brian ;-) It seems to be set up to reply to the sender (yuk).

Anyway, I've now moved it to the main list.

Cheers,

Ben.

> 
> rst

-- 
Ben Laurie                  Phone: +44 (181) 994 6435
Freelance Consultant and    Fax:   +44 (181) 994 6472
Technical Director          Email: ben@algroup.co.uk
A.L. Digital Ltd,           URL: http://www.algroup.co.uk
London, England.            Apache Group member (http://www.apache.org)

Mime
View raw message