You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Ruediger Pluem <rp...@apache.org> on 2007/02/05 21:26:25 UTC
Bumping question
Would the following change in mod_cache.h require a major bump or would a minor bump be enough?
Index: mod_cache.h
===================================================================
--- mod_cache.h (Revision 503593)
+++ mod_cache.h (Arbeitskopie)
@@ -314,7 +314,7 @@
APR_DECLARE_OPTIONAL_FN(apr_status_t,
ap_cache_generate_key,
- (request_rec *r, apr_pool_t*p, char**key ));
+ (request_rec *r, apr_pool_t*p, char**key, cache_request_rec *cache));
#endif /*MOD_CACHE_H*/
Or do we need no bump at all as we currently do not install mod_cache.h in the includes directory
and thus all things defined in mod_cache.h can be considered private?
Regards
Rüdiger
Re: Bumping question
Posted by Ruediger Pluem <rp...@apache.org>.
On 02/06/2007 04:16 AM, William A. Rowe, Jr. wrote:
> Ruediger Pluem wrote:
>
>>As 3rd parties might want to use the provider interface to provide their own storage
>>providers it might be worth discussing if we *should* make this a public API and should
>>install it via "make install".
>
>
> Someone probably has ;-) That said, +1 for this in trunk/httpd 2.4
> for certain. Because it looks like a moving target, and we are late
> to the game, I wouldn't really hassle with it for httpd 2.2.
Of course. I only wanted to make it a public API on trunk not on 2.2.
Regards
Rüdiger
Re: Bumping question
Posted by "William A. Rowe, Jr." <wr...@rowe-clan.net>.
Ruediger Pluem wrote:
>
> As 3rd parties might want to use the provider interface to provide their own storage
> providers it might be worth discussing if we *should* make this a public API and should
> install it via "make install".
Someone probably has ;-) That said, +1 for this in trunk/httpd 2.4
for certain. Because it looks like a moving target, and we are late
to the game, I wouldn't really hassle with it for httpd 2.2.
Re: Bumping question
Posted by Graham Leggett <mi...@sharp.fm>.
Ruediger Pluem wrote:
> As 3rd parties might want to use the provider interface to provide their own storage
> providers it might be worth discussing if we *should* make this a public API and should
> install it via "make install".
+1.
Regards,
Graham
--
Re: Bumping question
Posted by Ruediger Pluem <rp...@apache.org>.
On 02/06/2007 12:06 AM, Joe Orton wrote:
> On Mon, Feb 05, 2007 at 09:26:25PM +0100, Ruediger Pluem wrote:
>
>>Or do we need no bump at all as we currently do not install
>>mod_cache.h in the includes directory and thus all things defined in
>>mod_cache.h can be considered private?
>
>
> I vote definitely "no bump". If the header is not installed by "make
> install" it is by definition a private interface, and not part of the
> public API/ABI. So no need to worry about MMN bumps, nor maintaining
> backwards compat for 2.2.x, etc.
Thanks for the reply. I guess I got confused by myself as I did a minor bump about
a year ago in r370101 for adding further fields to the cache_server_conf struct
defined in mod_cache.h.
As 3rd parties might want to use the provider interface to provide their own storage
providers it might be worth discussing if we *should* make this a public API and should
install it via "make install".
Regards
Rüdiger
Re: Bumping question
Posted by Joe Orton <jo...@redhat.com>.
On Mon, Feb 05, 2007 at 09:26:25PM +0100, Ruediger Pluem wrote:
> Would the following change in mod_cache.h require a major bump or would a minor bump be enough?
...
> Or do we need no bump at all as we currently do not install
> mod_cache.h in the includes directory and thus all things defined in
> mod_cache.h can be considered private?
I vote definitely "no bump". If the header is not installed by "make
install" it is by definition a private interface, and not part of the
public API/ABI. So no need to worry about MMN bumps, nor maintaining
backwards compat for 2.2.x, etc.
joe
Re: Bumping question
Posted by "William A. Rowe, Jr." <wr...@rowe-clan.net>.
William A. Rowe, Jr. wrote:
> Ruediger Pluem wrote:
>> Would the following change in mod_cache.h require a major bump or would a minor bump be enough?
>>
>> APR_DECLARE_OPTIONAL_FN(apr_status_t,
>> ap_cache_generate_key,
>> - (request_rec *r, apr_pool_t*p, char**key ));
>> + (request_rec *r, apr_pool_t*p, char**key, cache_request_rec *cache));
>
> That one is a major bump by definition, you broke, you didn't enhance ABI.
FYI - if you created ap_cache_generate_key_ex for the lifetime of httpd 2.2,
you would solve the entire puzzle ;-)
Re: Bumping question
Posted by "William A. Rowe, Jr." <wr...@rowe-clan.net>.
Ruediger Pluem wrote:
> Would the following change in mod_cache.h require a major bump or would a minor bump be enough?
>
> APR_DECLARE_OPTIONAL_FN(apr_status_t,
> ap_cache_generate_key,
> - (request_rec *r, apr_pool_t*p, char**key ));
> + (request_rec *r, apr_pool_t*p, char**key, cache_request_rec *cache));
That one is a major bump by definition, you broke, you didn't enhance ABI.
> Or do we need no bump at all as we currently do not install mod_cache.h in the includes directory
> and thus all things defined in mod_cache.h can be considered private?
I'll agree with Joe on this, that since mod_cache.h isn't really a public
interface yet. But we better put a big red warning in CHANGES that we
broke any third party caching modules or providers.