Re: [git-for-windows] Git for Windows v2.15.0-rc prerelease, was Re: [ANNOUNCE] Git v2.15.0-rc2

[Johannes Schindelin <Johannes.Schindelin@gmx.de>]

Hi Lars,

On Fri, 27 Oct 2017, Lars Schneider wrote:

> > On 27 Oct 2017, at 14:11, Lars Schneider <larsxschneider@gmail.com> wrote:
> > 
> >> On 21 Oct 2017, at 00:22, Johannes Schindelin <johannes.schindelin@gmx.de> wrote:
> >> 
> >> [cutting linux-kernel]
> >> 
> >> On Fri, 20 Oct 2017, Junio C Hamano wrote:
> >> 
> >>> A release candidate Git v2.15.0-rc2 is now available for testing
> >>> at the usual places.
> >> 
> >> The Git for Windows equivalent is now available from
> >> 
> >>   https://github.com/git-for-windows/git/releases/tag/v2.15.0-rc2.windows.1
> > 
> > I just tested RC2 on Windows and I don't see my "Filtering content:"
> > output if I clone a Git repository with Git LFS files (and Git LFS
> > 2.3.3+ installed).
> > 
> > The feature was introduced in the following commit which is be part of
> > your RC2 build commit (b7f8941):
> > https://github.com/git/git/commit/52f1d62eb44faf569edca360ec9af9ddd4045fe0
> > 
> > On macOS everything works as expcted with RC2:
> >    ...
> >    remote: Total 15012 (delta 0), reused 0 (delta 0), pack-reused 15012
> >    Receiving objects: 100% (15012/15012), 2.02 MiB | 753.00 KiB/s, done.
> >    Filtering content:  43% (6468/15000), 33.30 KiB | 0 bytes/s
> >    ...
> > 
> > Do you, or other Windows experts, spot something in the commit linked
> > above that could cause trouble on Windows?
> 
> Well, it turns out the output works for my real life repos but not for
> my Git LFS testing repo.
> 
>     git clone https://github.com/larsxschneider/lfstest-manyfiles
> 
> ... prints the filtering content output on macOS but not on Windows.
> The progress function has some delay feature that suppresses the output
> if it is only shown for a second or something. However, in this test case
> the output should be visible for several seconds at least...
> I am still puzzled.

Nothing really strikes me as obvious.

Do you do this in Git Bash? If so, maybe you can also test in Git CMD? I
do remember having issues with stderr only showing up in time if it was
fflush()ed explicitly, but only in Git Bash (i.e. a MinTTY problem).

Ciao,
Dscho