Re: [lmi] PATCH: Switch to using std::filesystem

[Vadim Zeitlin]

On Thu, 29 Apr 2021 02:33:07 +0000 Greg Chicares <gchicares@sbcglobal.net> 
wrote:

GC> On 4/26/21 9:53 PM, Vadim Zeitlin wrote:
GC> [...]
GC> >  The rest of the changes just deal with the differences between the two
GC> > libraries. They're done in different commits to make it simpler to review
GC> > them, but should, for once, be applied all together, i.e. the entire PR
GC> > should be "squash merged", as almost all of these commits are needed for
GC> > the code to actually compile and work, and preserving their history is not
GC> > very important.
GC> 
GC> OTOH, is squashing very important?

 No, not really, but I just wonder why wouldn't you want to do it? Normally
it should be as simple as "git merge --squash xanadu/std-fs" followed by
"git commit".

GC> Do you have a strong
GC> preference for me to squash this into a single commit?
GC> Would you object if it enters origin/master as thirty-four
GC> distinct commits?

 Not "strong" and not really an objection, but this would create a problem
if anyone wants to run "git bisect" in the future, as ~30 commits among
those 34 wouldn't build, so would have to be skipped when bisecting. It's
not a huge deal, but OTOH it does seem a bit wrong to have non-buildable
commits in master (they bother me less in a branch and you can just mark
the branch fork point and merge point as good for git-bisect to realize
that it doesn't need to go into it at all).

 So, again, please do it like this if you have to, but I'm still curious
why do you think you do -- please let me know if you have a moment, as it
might indicate some other problem I'm not aware of.

 Thanks,
VZ

pgpx27g9WNLo4.pgp
Description: PGP signature