More function naming conflicts

[Rasmus Lerdorf <ra...@lerdorf.on.ca>]

Somebody somewhere is really out to get me.  The latest Solid (an RDBMS)
libs have no less than 5 function naming conflicts with Apache.  I don't
see why in the world they have a "send_http_header" function for example,
but something like "process request" is way too generic a name for Apache
to have as a non-static function.  It would be nice if every single
non-static Apache function was clearly labelled with whatever prefix it
was we decided on for these things.

http_request.o: Definition of symbol `process_request' (multiply defined)
http_protocol.o: Definition of symbol `send_http_header' (multiply defined)
util.o: Definition of symbol `ht_time' (multiply defined)
util.o: Definition of symbol `gm_timestr_822' (multiply defined)
util.o: Definition of symbol `unescape_url' (multiply defined)

-Rasmus

Re: More function naming conflicts

[Rasmus Lerdorf <ra...@lerdorf.on.ca>]

> Hmmph.  I don't see why *they* can't label *their* functions..

Yes, obviously they should prefix their functions as well.  And I have
complained to them about it too.  Apache is no less at fault though.

> Oh, well.  The question's open again.  We've got these, and palloc(),
> and wasn't there another one as well?  Let's have another round of
> discussion on temporary (1.3) prefixes, shall we?

new_connection() was the other one which conflicts with the IMAP libs.

-Rasmus

Re: More function naming conflicts

[Rodent of Unusual Size <Ke...@Golux.Com>]

Rasmus Lerdorf wrote:
> 
> Somebody somewhere is really out to get me.  The latest Solid (an RDBMS)
> libs have no less than 5 function naming conflicts with Apache.  I don't
> see why in the world they have a "send_http_header" function for example,
> but something like "process request" is way too generic a name for Apache
> to have as a non-static function.  It would be nice if every single
> non-static Apache function was clearly labelled with whatever prefix it
> was we decided on for these things.

Hmmph.  I don't see why *they* can't label *their* functions..

Oh, well.  The question's open again.  We've got these, and palloc(),
and wasn't there another one as well?  Let's have another round of
discussion on temporary (1.3) prefixes, shall we?

#ken	P-)}

Re: More function naming conflicts

[Marc Slemko <ma...@worldgate.com>]

On Fri, 6 Feb 1998, Dean Gaudet wrote:

> I wonder if they've linked in a restriced Apache or NCSA server.  That
> would be one explanation for this.  Any chance they left static symbols

Then gee, we would have an awful hard time renaming functions to avoid
that.  <g>

We really need a language that offers easy multiple scopes.  So we can say
that some functions aren't in the API but can be used by "closely coupled"
modules that need to go beyond the regular API, then we could also have a
strict list of functions defined to be the API and normal modules could
only use them.

Sigh.  

> names in the binary as well? 
> 
> Dean
> 
> On Fri, 6 Feb 1998, Rasmus Lerdorf wrote:
> 
> > Somebody somewhere is really out to get me.  The latest Solid (an RDBMS)
> > libs have no less than 5 function naming conflicts with Apache.  I don't
> > see why in the world they have a "send_http_header" function for example,
> > but something like "process request" is way too generic a name for Apache
> > to have as a non-static function.  It would be nice if every single
> > non-static Apache function was clearly labelled with whatever prefix it
> > was we decided on for these things.
> > 
> > http_request.o: Definition of symbol `process_request' (multiply defined)
> > http_protocol.o: Definition of symbol `send_http_header' (multiply defined)
> > util.o: Definition of symbol `ht_time' (multiply defined)
> > util.o: Definition of symbol `gm_timestr_822' (multiply defined)
> > util.o: Definition of symbol `unescape_url' (multiply defined)
> > 
> > -Rasmus
> > 
> > 
> 

Re: More function naming conflicts

[Dean Gaudet <dg...@arctic.org>]

I wonder if they've linked in a restriced Apache or NCSA server.  That
would be one explanation for this.  Any chance they left static symbols
names in the binary as well? 

Dean

On Fri, 6 Feb 1998, Rasmus Lerdorf wrote:

> Somebody somewhere is really out to get me.  The latest Solid (an RDBMS)
> libs have no less than 5 function naming conflicts with Apache.  I don't
> see why in the world they have a "send_http_header" function for example,
> but something like "process request" is way too generic a name for Apache
> to have as a non-static function.  It would be nice if every single
> non-static Apache function was clearly labelled with whatever prefix it
> was we decided on for these things.
> 
> http_request.o: Definition of symbol `process_request' (multiply defined)
> http_protocol.o: Definition of symbol `send_http_header' (multiply defined)
> util.o: Definition of symbol `ht_time' (multiply defined)
> util.o: Definition of symbol `gm_timestr_822' (multiply defined)
> util.o: Definition of symbol `unescape_url' (multiply defined)
> 
> -Rasmus
> 
>