lmi
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [lmi] PATCH: C++20 and clang


From: Greg Chicares
Subject: Re: [lmi] PATCH: C++20 and clang
Date: Tue, 20 Apr 2021 00:38:45 +0000
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101 Thunderbird/78.9.0

On 4/19/21 9:36 PM, Vadim Zeitlin wrote:
> On Mon, 19 Apr 2021 21:19:26 +0000 Greg Chicares <gchicares@sbcglobal.net> 
> wrote:
[...]
> GC> $git log -10 --oneline xanadu/clang-cxx20
> GC> e1ffd97fe (xanadu/clang-cxx20) Disable clang warnings about using 
> volatile in C++20 mode
> GC> 9ce718859 Disable -Wunused-parameter warning in Boost.Numeric header
> GC> 1a86b8f35 Use std::endian when using C++20
> GC> e4cb44b01 Disable clang enum-related warnings in C++20 mode
> GC> ae672dce6 Use wxSizerFlags for creating the "About" dialog
> GC> 9a4888100 Use constexpr instead of enum for defining constants
> GC> 709aa9fa0 Enable verbose build in CI autotools jobs
> GC> 54cd54b6f Don't show test-suite.log in CI script if it doesn't exist
> GC> e54575c99 Use C++20 when building lmi with autotools
> GC> 8be683cc1 Record speed measurements
> 
>  This is surprising, the commits up to and including 709aa9fa0 are already
> part of the main repository at Savannah because I had pushed them there
> (they only affect "my" files). In fact, there is another commit that I
> pushed since then (9414edf44ff43ee28a28d2b1efe2abd5d8dfa1fa) and I don't
> understand why didn't your "git-fetch --all" fetch it from there. Surely
> you have a configured remove (probably called "origin"?) for the Savannah
> repository?

Indeed I do have 'origin' configured to point to savannah.
But I hadn't pulled from savannah since yesterday, so I
was probably comparing your branch to local master.

I believe I've merged everything now. I did some fixups
to make 'make $coefficiency check_concinnity' succeed.

>  In any case, the commits that are part of this PR are the first 6 commits
> in the list above (they're also those that appear on the web page if you
> open the URL above in the browser).
>From your previous messages, it seems that you took extra
pains to present this patchset in various ways to help me.
That's appreciated, but it's not really necessary as I
don't take advantage of all the diversity. Let me explain
what I did here, because it may save you time in future.

My session looks like this:

$git --no-pager log -6 --oneline xanadu/clang-cxx20 origin/master
e1ffd97fe (xanadu/clang-cxx20) Disable clang warnings about using volatile in 
C++20 mode
9ce718859 Disable -Wunused-parameter warning in Boost.Numeric header
1a86b8f35 Use std::endian when using C++20
e4cb44b01 Disable clang enum-related warnings in C++20 mode
ae672dce6 Use wxSizerFlags for creating the "About" dialog
9414edf44 (xanadu/master, origin/master, origin/HEAD) Use "make --keep-going" 
in autotools CI jobs

[I can scroll back to that, to see what to apply next.]

$git cherry-pick ae672dce6                                       

$git cherry-pick e4cb44b01

and so on, through e1ffd97fe. Thus, what I did was:
 - cherry-pick one commit at a time (oldest first)
 - locally examine and test
 - [make some notes about any later fixups]
 - repeat until all are applied
and then go back and use git-rebase for fixups; and,
finally, run a comprehensive test before pushing.

I read the commit messages, but not the github website.
The only reason I look at github is to find the name
of the branch (to map PR#179 to "clang-cxx20"). I'd
rather study a branch with tools of my own choosing
than with github's.

For a different patchset, I might first cherry-pick an
entire series of commits to see the overall net effect,
and then perhaps accept them all, though at other times
I might then 'git reset --hard origin/master' and go back
and cherry-pick them individually. But I really use github
only as a git server, not as a user interface.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]