[haiku-development] Re: Moving away from Subversion (pt 4)
- From: Niels Sascha Reedijk <niels.reedijk@xxxxxxxxx>
- To: haiku-development@xxxxxxxxxxxxx
- Date: Mon, 6 Dec 2010 18:47:06 +0100
Hi,
On 6 December 2010 18:36, Adrien Destugues <adestugu@xxxxxxxxx> wrote:
>
>> Not intending to troll here, but why exactly do you need readable
>> version numbers for a DVCS? Are you going to speak them over the
>> phone?
>
> Sequential version numbers are the easy way to binary search which
> commit broke some particular feature. I guess some date tag would also
> work, but splitting
> in the middle of the search space is just easier with sequential
> numbers.
Though I fully understand the critique that these revision numbers are
useless as soon as you use the decentralized revision control system,
you know, decentralized. A local commit means that all the revisions
that follow are re-ordered locally, meaning basically that the uses of
revision numbers in communication with others is limited.
Furthermore, when the tool is used properly you get what is known as
branchy development, which means that you will have revision 5, 6, 7
and 8, but 6 and 7 might have never been on the mainline (they were
merged in as 8). Only bzr does this kind of numbering in a way that is
sane, by making the numbers depend on the branch they are in. In that
way, sequential numbers do refer to a single (mainline) branch.
This does not have to be a problem for the binary searching use case
though. The only thing is that in communicating the broken changeset,
the changeset id is probably a better communication tool.
Regards,
N>
Other related posts:
