Re: [PATCH] clean: explicitly `fflush` stdout

From: Jeff King <peff@peff.net>
Date: Apr 8, 2020, 20:14 UTC
Message-ID: <20200408201454.GB2270445@coredump.intra.peff.net>
In-Reply-To: <pull.755.git.git.1586374380709.gitgitgadget@gmail.com>

On Wed, Apr 08, 2020 at 07:33:00PM +0000, Johannes Schindelin via GitGitGadget wrote:

> From: =?UTF-8?q?=EB=A7=88=EB=88=84=EC=97=98?= <nalla@hamal.uberspace.de>
> 
> For performance reasons `stdout` is buffered by default. That leads to
> problems if after printing to `stdout` a read on `stdin` is performed.

This first paragraph left me scratching my head. It's only a problem for interactive uses, right? Would:

  This can lead to problems for interactive commands which write a
  prompt to `stdout` and then read user input on `stdin`. If the prompt
  is left in the buffer, the user will not realize the program is
  waiting for their input.

make sense?

> For that reason interactive commands like `git clean -i` do not function
> properly anymore if the `stdout` is not flushed by `fflush(stdout)` before
> trying to read from `stdin`.

I'm not sure I understand this "anymore". Did the buffering change at some point, introducing a regression?

> So let's precede all reads on `stdin` in `git clean -i` by flushing
> `stdout`.

This (and the patch) make sense to me. It might be worth factoring out a "read input from user" function and putting the flush there, but with only three affected call sites, I'm OK with it either way.

>     This is yet another patch that was funneled through a Git for Windows
>     PR. It has served us well for almost five years now, and it is beyond
>     time that it find its final home in core Git.

Agreed. I guess it didn't affect people on other platforms much because stdout to a terminal is generally line-buffered on POSIX systems. But it's better not to rely on that, plus it could create confusion if somebody tried to manipulate the interactive operation through a pipe (e.g., driving it from a GUI or something).

-Peff