On Mon, Jan 21, 2019 at 3:18 PM Craig Ozancin <[email protected]> wrote:
> I have thought the same thing. Absorb has been a bit of a hit/miss for me. > Absorb is very impressive, but all to often there are changes that it does > not identify for a specific revision and I end up using histedit to get > them in place. > Surely you know this, but for the benefit of other readers: The problem with ‘histedit’ in this particular scenario is that you may end up wasting time reviewing merges for no reason. The algorithm used by ‘absorb’ appears a lot more focused on the actual change. Only today I noticed in a series of interactive micro-commits that I forgot to add an entire function definition. No way a language unaware tool could guess that surely this definition should be added to the same changeset which added the corresponding calls, leaving ‘absorb’ by itself totally clueless. I ended up using ‘histedit’, which brought up the merge tool, suggesting that the merge may be more complicated than expected. It wasn't in the end but making sure cost me quite a bit of time in reviewing this. The whole process cost me two merge reviews. By specifying an explicit target revision, I guess (hope, actually) ‘absorb’ help in a use case that I think may not be that uncommon after all. Cheers, Marcus
_______________________________________________ Mercurial mailing list [email protected] https://www.mercurial-scm.org/mailman/listinfo/mercurial
