Hi Martin,

On Thu, 9 Apr 2020, Martin Ågren wrote:

> On Thu, 9 Apr 2020 at 12:26, Johannes Schindelin via GitGitGadget
> <gitgitgadget@gmail.com> wrote:
>
> >         handle = CreateFileW(wfilename, FILE_APPEND_DATA,
> >                         FILE_SHARE_WRITE | FILE_SHARE_READ,
> >                         NULL, create, FILE_ATTRIBUTE_NORMAL, NULL);
> > -       if (handle == INVALID_HANDLE_VALUE)
> > -               return errno = err_win_to_posix(GetLastError()), -1;
>
> (This code, which is being removed, used "," to avoid having to introduce
> braces. There's another instance of this pattern a bit above.)

Oh yes, that's an important point I had missed.

> > +       if (handle == INVALID_HANDLE_VALUE) {
>
> (Adding a brace here.)
>
> > +               DWORD err = GetLastError();
> > +               /*
> > +                * Some network storage solutions (e.g. Isilon) might return
> > +                * ERROR_INVALID_PARAMETER instead of expected error
> > +                * ERROR_PATH_NOT_FOUND, which results in a unknow error. If
>
> s/ a / an /
> s/unknow/&n/

Oh shucks. Usually I try to do a final review of the patches before I
"upstream" them. Seems like the quality of my submission here is not what
I want it to be.

>
> > +                * so, the error is now forced to be an ERROR_PATH_NOT_FOUND
> > +                * error instead.
> > +                */
>
> "is now forced" sounds more like it describes this change/commit, rather
> than this piece of code. Maybe this final sentence can be scrapped
> entirely, since the forcing/translating/mapping is obvious from the code
> anyway? The remainder of the comment goes into *why* and looks more
> useful. Just my 2 cents.
>
> > +               if (err == ERROR_INVALID_PARAMETER)
> > +                       err = ERROR_PATH_NOT_FOUND;
> > +               return errno = err_win_to_posix(err), -1;
> > +       }
>
> Now there's no need to avoid introducing braces, so maybe split this
> into two lines for a lower huh-factor?
>
>   errno = err_win_to_posix(err);
>   return -1;

Absolutely.

I updated the PR at https://github.com/git/git/pull/756 (also addressing
issues with the commit message) and will wait for while before sending v2.

Thank you for your excellent review,
Dscho