httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
Subject Re: [PATCH] APR wrapper for iconv
Date Wed, 19 Apr 2000 00:27:40 GMT

Obviously I have been really unclear today, so I'll try again.

Jeff submitted patches to deal with part 1 of a two part solution.
Through comments, the second part was brought up.  The second part is NOT
EBCDIC-> ASCII or ASCII->EBCDIC.  The first part translates between ASCII
And EBCDIC going in either direction.  The second part is to convert
between two ASCII codepages.

Jeff's changes started implementing ECBDIC->ASCII and back.  I had a few
minor issues, which I made, and AFAICT, JEff has read, understood, and
taken them into account for the actual patch. After I made comments, about
five other people have interpretted what I said.

My issues were:
   getting the name change right (I said I could live with the name
   macros so that are essentially no-ops (evaluate to APR_ENOTIMPL) for
          platforms that don't do codepage translation.

I then asked that we focus on the first step (the code Jeff has already
indicated interest in doing!), and ignore the second step until later.  I
also expressed that I would have no problem leaving the second step until
after 2.0 was released, because it is a very new feature, and the first
step is important for 2.0, because it is a current feature of Apache 1.3.

I am not saying Jeff can't do something, other than to say if you are
going to do the whole thing, please let us know so that we can hack
something in for 2.0.  I am not saying you must leave any part of this
until 2.0 is released (although my vote is to leave part of it until 2.1
of Apache and 1.1 of APR).  All I am saying is "realize this is two parts
and please treat it as two parts."  They are similar, but not the same.
They should be coded in two steps, so that we can get the problems ironed
out with the first part, and then incorporate the second part.

I am leaving this conversation, because amazingly enough, the only people
that are currently in this conversation are arguing about code that none
of us plans to implement.

Jeff, I hope I have now made my position clear, if you have any more
questions about my issues with original patch, please ask me.  :-)


On Tue, 18 Apr 2000, Greg Stein wrote:

> On Wed, 19 Apr 2000, David Reid wrote:
> > > > Agreed -- dividing the problem is usually The Right
> > > > Thing(tm).  [...]
> > > > That still leaves the EBCDIC work to be done by somebody [...]
> > >
> > > Or, by some small miracle, the sum and total of Jeff's solution
> > > may provide that implementor with a very quick and easy
> > > ascii->ebcdic->ascii solution :-)
> > >
> > > Implement the complex function and discover if it solves
> > > this simple subset of the issue.
> > 
> > I think Ryan is just concerned that we bite off too much now and in a couple
> > of months we still don't have EBCDIC->ascii working.  If Jeff can get it all
> > working in a few days then great, but if he can't...
> <meta-issue-rant>
> And what if Jeff wants to work on just this one issue for the next six
> months and NOT work on the EBCDIC issue?
> Who are we to say "no. you can't do that." ??
> We know that we'd like the codepage translation stuff in the long run. Why
> demand it must be long, if somebody wants to do short? What happened to
> the credo that people can work on the thing that they *like* to work on?
> Sure, if somebody wants to add an IMAP server module to Apache, then we
> might take issue (but maybe not). But that isn't the case here.
> </meta-issue-rant>
> Cheers,
> -g
> -- 
> Greg Stein,

Ryan Bloom               
406 29th St.
San Francisco, CA 94131

View raw message