[sr-dev] git:master: core and several modules: instance and reg_id in branch_t

[Daniel-Constantin Mierla]

On 12/8/12 1:55 PM, Juha Heinanen wrote:
> Daniel-Constantin Mierla writes:
>
>> You can commit locally and push all at once, so practically all of them
>> will be available on remote repository at the same time. I did it many
>> times (it is actually the usual way because it easier to review later or
>> send only specific links to the authors of the components I want to be
>> reviewed).
> this was all or nothing change.  making several commit locally would
> just meant more work for me.

I think this is not really a standing argument in the overall context. 
It is very important to keep robustness of the application, meaning that 
when one affects the components of other developers, it has to be an 
easy way for them to spot the changes.

It is just about running several commit commands instead of one, the 
push to remote is one command. What would change is the subject saying 
what component was affected. If there is no proper time at the moment 
for something, better don't rush it, could be harder later to sort it out.

>
> rather than argue about administrative things,

I don't think this is an administrative thing, it is related to 
development. Again, it was a suggestion. I did it because it happened to 
bring in an invalid change. It was not an enforcement. Looking at very 
large commits require also lot of time, usually has to be done in one 
time frame. Smaller commits can be reviewed easier sequentially, at 
different points in time.


>   i would like to get
> comments on this, because it needs to be done one way or the other next:

Well, the master branch doesn't compile, pv module throws errors.

Also, the modules in modules_s have to be made to compile for the 
moment, otherwise master branch compilation fails anyhow. They may be 
removed, but that will still take time, it will not happen in the next 
hours.

>
>> if we do that, then for consistency, get_branch and next_branch should
>> be dropped altogether and replaced by get_sip_branch (which already
>> exists) and get_next_sip_branch (which would be new).
> so either new arguments to get_branch and next_branch or replace them
> with get_sip_branch and get_next_sip_branch.
It is no really need to replace them (the old ones). You can add a new 
function returning the pointer to the next branch and use it. For long 
term, I think the one with the pointer is easier to maintain whenever we 
need to add new members in the structure, I would go for that.

However, the old ones can be kept, so not much code is affected, 
probably not all places where get_branch/next_branch are used need the 
instance/reg-id.

Cheers,
Daniel

-- 
Daniel-Constantin Mierla - http://www.asipto.com
http://twitter.com/#!/miconda - http://www.linkedin.com/in/miconda