You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Ben Laurie <be...@algroup.co.uk> on 2003/05/03 16:10:21 UTC

derived files in CVS?

Why are modules/ssl/ssl_expr_parse.c, modules/ssl/ssl_expr_parse.h and
modules/ssl/ssl_expr_scan.c in CVS? They're all generated from other files.

Cheers,

Ben.

-- 
http://www.apache-ssl.org/ben.html       http://www.thebunker.net/

"There is no limit to what a man can do or how far he can go if he
doesn't mind who gets the credit." - Robert Woodruff


Re: derived files in CVS?

Posted by Ben Laurie <be...@algroup.co.uk>.
Justin Erenkrantz wrote:

> --On Saturday, May 3, 2003 7:41 PM +0100 Ben Laurie <be...@algroup.co.uk>
> wrote:
> 
>> Of course it is! Our stance has always been that developers have to have
>> _all_ the tools available.
> 
> 
> IIRC, the issue came up was that different versions of flex/bison
> (whatever) produce very different code (and in some special cases, very
> broken). Therefore, it isn't very reliable.  Since this is mainline code
> (unlike configure which has no direct impact on the resulting binary), I
> believe the intent was to make the output well-known and tested.

I think we can simply tell developers which versions are OK and expect
them to install them, no? I find this a little unbelievable, though...

> I believe PHP jumps through a lot of hoops to verify the output of flex
> and bison and that only helps to a limited degree (they've still been
> burned). Someone more familiar with PHP might be able to shed some
> light.  -- justin

Don't forget we're _not_ talking about the distro here, but I would love
to hear from the PHP folk, indeed.

Cheers,

Ben.

-- 
http://www.apache-ssl.org/ben.html       http://www.thebunker.net/

"There is no limit to what a man can do or how far he can go if he
doesn't mind who gets the credit." - Robert Woodruff


Re: derived files in CVS?

Posted by Justin Erenkrantz <ju...@erenkrantz.com>.
--On Saturday, May 3, 2003 7:41 PM +0100 Ben Laurie <be...@algroup.co.uk> wrote:

> Of course it is! Our stance has always been that developers have to have
> _all_ the tools available.

IIRC, the issue came up was that different versions of flex/bison (whatever) 
produce very different code (and in some special cases, very broken). 
Therefore, it isn't very reliable.  Since this is mainline code (unlike 
configure which has no direct impact on the resulting binary), I believe the 
intent was to make the output well-known and tested.

I believe PHP jumps through a lot of hoops to verify the output of flex and 
bison and that only helps to a limited degree (they've still been burned). 
Someone more familiar with PHP might be able to shed some light.  -- justin

Re: derived files in CVS?

Posted by Ben Laurie <be...@algroup.co.uk>.
Jeff Trawick wrote:

> William A. Rowe, Jr. wrote:
> 
>> At 09:10 AM 5/3/2003, Ben Laurie wrote:
>>
>>> Why are modules/ssl/ssl_expr_parse.c, modules/ssl/ssl_expr_parse.h and
>>> modules/ssl/ssl_expr_scan.c in CVS? They're all generated from other
>>> files.
>>
>>
>>
>> Because we don't expect the user to have bison/lex/yacc handy.
> 
> 
> We don't put configure in CVS either; the person rolling the tarball has
> to have the right tools on their machine to generate it.
> 
> Is it reasonable to do the same for the SSL scanner/parser?  It is a bit
> of a drag when they're regenerated locally and then dominate cvs diff
> output.

Of course it is! Our stance has always been that developers have to have
_all_ the tools available.

Cheers,

Ben.

-- 
http://www.apache-ssl.org/ben.html       http://www.thebunker.net/

"There is no limit to what a man can do or how far he can go if he
doesn't mind who gets the credit." - Robert Woodruff


Re: derived files in CVS?

Posted by Jeff Trawick <tr...@attglobal.net>.
William A. Rowe, Jr. wrote:
> At 09:10 AM 5/3/2003, Ben Laurie wrote:
> 
>>Why are modules/ssl/ssl_expr_parse.c, modules/ssl/ssl_expr_parse.h and
>>modules/ssl/ssl_expr_scan.c in CVS? They're all generated from other files.
> 
> 
> Because we don't expect the user to have bison/lex/yacc handy.

We don't put configure in CVS either; the person rolling the tarball has 
to have the right tools on their machine to generate it.

Is it reasonable to do the same for the SSL scanner/parser?  It is a bit 
of a drag when they're regenerated locally and then dominate cvs diff 
output.



Re: derived files in CVS?

Posted by "William A. Rowe, Jr." <wr...@rowe-clan.net>.
At 09:10 AM 5/3/2003, Ben Laurie wrote:
>Why are modules/ssl/ssl_expr_parse.c, modules/ssl/ssl_expr_parse.h and
>modules/ssl/ssl_expr_scan.c in CVS? They're all generated from other files.

Because we don't expect the user to have bison/lex/yacc handy.