Would be nice to see some detailed explanation about cases when developer must 
use extended version fields, with examples: i.e. open some "feature" issue and 
explain your suggestions
there.

Alexander

On 21.08.2013, at 23:22, Alexander Pyhalov <[email protected]> wrote:

> Alexander Eremin писал 21.08.2013 21:25:
>> IPS pkg version has enough fields (four) for this. May be for some
>> cases just use third field (branch version), which provides
>> vendor-specific info, can contain build number or any other info?
> 
> Yes, I don't suggest change IPS, just userland makefiles which generate 
> version numbers - it still can be encoded in first and third field. I think 
> we just need some agreements on modified versioning scheme and perhaps some 
> tools to enforce this scheme. For exampel IPS_COMPONENT_SUBVERSION can be the 
> last number in third field.
> 
>> On 21.08.2013, at 21:12, Alexander Pyhalov <[email protected]> wrote:
>> Hello.
>> I'd like to suggest the following topic for discussion - package versions. 
>> We have several problems in /hipster:
>> a) to trigger package update it is necessary to touch Makefile
>> b) we need an indication that package wasn't just rebuilt, but was changed
>> c) we don't have a rollback mechanism for package updates.
>> There are several possible actions:
>> 1) add IPS_COMPONENT_SUBVERSION variable which indicates  local change of 
>> the package and must be changed with every package modification, so that
>> package number looks like 
>> $(IPS_COMPONENT_VERSION).$(IPS_COMPONENT_SUBVERSION),$(BUILD_VERSION) or 
>> similar;
>> 2) for packages having spoiled package versions introduce var like 
>> IPS_FICTIVE_VERSION with arbitrary high values (100, 1000, etc), set 
>> IPS_COMPONENT_VERSION to it, IPS_HUMAN_VERSION to real version and use 
>> incremented IPS_COMPONENT_SUBVERSION.
> 
> ---
> System Administrator of Southern Federal University Computer Center
> 
> 
> 
> _______________________________________________
> oi-dev mailing list
> [email protected]
> http://openindiana.org/mailman/listinfo/oi-dev

_______________________________________________
oi-dev mailing list
[email protected]
http://openindiana.org/mailman/listinfo/oi-dev

Reply via email to