[Scons-dev] Scons 2.3.2 regression, D tool...

Russel Winder russel at winder.org.uk
Sun Aug 10 05:19:29 EDT 2014


On Sat, 2014-08-09 at 22:41 +0200, Dirk Bächle wrote:
[…]
> I may have read Russel's comment wrong, and understood it like "The 
> result is messy, so we don't have a proper workflow and need to switch 
> the VCS.". This would be wrong in my opinion..but this meaning probably 
> wasn't intended anyway.

Slightly wrong, it was OK up to workflow and then diverged from
reality ;-)

My point was that maintaining a long running published feature clone is
fine, but it cannot be used as a base for a pull request without
creating an enormous mess on merging. I wasn't entirely worried by this,
Gary was. I therefore created a single changeset for the entire diff and
Gary merged this. In hindsight this was almost certainly not the right
thing to do since it led to a total loss of all history and a huge
changeset with lots of things going on. I wonder if the compromise is to
rebase to create a not entirely accurate history per se but all the
changesets as committed kept. The only danger is introducing new merge
conflicts during the rebasing.

> About "information": I'm a fan of keeping history as it 
> happens...including errors, and their immediate fixes. Today, I don't 
> know what information I might need tomorrow, or in 2 weeks/months. The 
> seemingly unusable commit (or even only its checkin comment) that I 
> rebase/squash away today, might serve as a starting point for a fresh 
> approach at a later time. Highly speculative, I know...but that's how I 
> see it.

I am a fan of history as it was, hence not using rebase in Bazaar or
Mercurial, and only using it in Git working as a client to Subversion
repoistories. However this is clearly inconsistent with having a nice
mainline history, and so the compromise appears to be all pull requests
coming from rebased feature clones.

The problem then is that pull requests need to be processed very quickly
to preserve the immediacy of the changeset. I am not sure we are set up
for this, so it isn't an obviously successful strategy.

There must be a compromise process in here somewhere…


-- 
Russel.
=============================================================================
Dr Russel Winder      t: +44 20 7585 2200   voip: sip:russel.winder at ekiga.net
41 Buckmaster Road    m: +44 7770 465 077   xmpp: russel at winder.org.uk
London SW11 1EN, UK   w: www.russel.org.uk  skype: russel_winder
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 181 bytes
Desc: This is a digitally signed message part
URL: <http://two.pairlist.net/pipermail/scons-dev/attachments/20140810/00a6adfc/attachment-0001.pgp>


More information about the Scons-dev mailing list