From: Johannes Schindelin <Johannes.Schindelin@gmx.de>
To: Junio C Hamano <gitster@pobox.com>
Cc: git@vger.kernel.org, Eric Sunshine <sunshine@sunshineco.com>
Subject: Re: [PATCH v2 1/7] log-tree: respect diffopt's configured output file stream
Date: Tue, 21 Jun 2016 09:31:45 +0200 (CEST) [thread overview]
Message-ID: <alpine.DEB.2.20.1606210920510.22630@virtualbox> (raw)
In-Reply-To: <xmqqwplju74a.fsf@gitster.mtv.corp.google.com>
Hi Junio,
On Mon, 20 Jun 2016, Junio C Hamano wrote:
> Johannes Schindelin <johannes.schindelin@gmx.de> writes:
>
> > The diff options already know how to print the output anywhere else
> > than stdout. The same is needed for log output in general, e.g. when
> > writing patches to files in `git format-patch`. Let's allow users to
> > use log_tree_commit() *without* changing global state via freopen().
>
> I wonder if this change is actually fixing existing bugs. Are there
> cases where diffopt.file is set, i.e. the user expects the output to be
> sent elsewhere, but the code here unconditionally emits to the standard
> output? I suspect that such a bug can be demonstratable in a test or
> two, if that were the case.
It is conceivable, but I did not have time to chase those cases down yet.
> I am sort-of surprised that we didn't do this already even though we
> had diffopt.file for a long time since c0c77734 (Write diff output
> to a file in struct diff_options, 2008-03-09).
>
> Use of freopen() to always write patches through stdout may have
> been done as a lazy workaround of the issue this patch fixes, but
> what is surprising to me is that doing it the right way like this
> patch does is not that much of work. Perhaps that was done long
> before c0c77734 was done, which would mean doing it the right way
> back then when we started using freopen() it would have been a lot
> more work and we thought taking a short-cut was warranted.
Back when I implemented the feature to write to individual files, I indeed
used freopen() out of laziness: 0377db7 (Teach fmt-patch to write
individual files., 2006-05-05). I could not have used diffopt.file at that
stage, anyway: that field still was almost two years in the future.
> > if (opt->children.name)
> > show_children(opt, commit, abbrev_commit);
> > show_decorations(opt, commit);
> > if (opt->graph && !graph_is_commit_finished(opt->graph)) {
> > - putchar('\n');
> > + fputc('\n', opt->diffopt.file);
>
> Hmph. putc() is the "to the given stream" equivalent of putchar()
> in the "send to stdout" world, not fputc(). I do not see a reason
> to force the call to go to a function avoiding a possible macro here.
TBH I did not even *know* putc(). It is amazing how you can learn new
things about the POSIX API after decades of working with it.
> Likewise for all the new fputc() calls in this series that were
> originally putchar().
Goes without saying.
> > @@ -880,8 +880,9 @@ int log_tree_commit(struct rev_info *opt, struct commit *commit)
> > shown = 1;
> > }
> > if (opt->track_linear && !opt->linear && opt->reverse_output_stage)
> > - printf("\n%s\n", opt->break_bar);
> > + fprintf(opt->diffopt.file, "\n%s\n", opt->break_bar);
> > opt->loginfo = NULL;
> > - maybe_flush_or_die(stdout, "stdout");
> > + if (opt->diffopt.file == stdout)
> > + maybe_flush_or_die(stdout, "stdout");
> > return shown;
> > }
>
> This one looks fishy.
>
> Back when we freopen()'ed to write patches only through stdout, we
> always called maybe_flush_or_die() to make sure that the output is
> flushed correctly after processing each commit. This change makes
> it not to care, which I doubt was what you intended. Instead, my
> suspicion is that you didn't want to say "stdout" when writing into
> a file.
>
> But even when writing to on-disk files, the code before your series
> would have said "stdout" when it had trouble flushing, so I do not
> think this new "if()" condition is making things better. If "it
> said stdout when having trouble flushing to a file" were a problem
> to be fixed, "let's not say stdout by not even attempting to flush
> and catch errors when writing to a file" would not be the right
> solution, no?
>
> Personally, I do not think it hurts if we kept saying 'stdout' here,
> even when we flush opt->diffopt.file and found a problem.
Okay, I changed it back to be unconditional.
My original thinking was that we will fclose() the file (if it is not
stdout) anyway, which implies flushing. But the more I think about it,
the more I come to the conclusion that this is more of a side effect,
based on deep knowledge of the current code. So I now agree with you that
it would be "too clever".
Expect v3 in a moment.
Ciao,
Dscho
next prev parent reply other threads:[~2016-06-21 7:32 UTC|newest]
Thread overview: 67+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-06-18 10:03 [PATCH 0/5] Let log-tree and friends respect diffopt's `file` field Johannes Schindelin
2016-06-18 10:03 ` [PATCH 1/5] log-tree: respect diffopt's configured output file stream Johannes Schindelin
2016-06-18 10:03 ` [PATCH 2/5] line-log: " Johannes Schindelin
2016-06-18 10:04 ` [PATCH 3/5] graph: respect the diffopt.file setting Johannes Schindelin
2016-06-18 10:04 ` [PATCH 4/5] shortlog: support outputting to streams other than stdout Johannes Schindelin
2016-06-18 10:04 ` [PATCH 5/5] format-patch: avoid freopen() Johannes Schindelin
2016-06-19 20:01 ` Eric Sunshine
2016-06-20 6:26 ` Johannes Schindelin
2016-06-20 6:32 ` Eric Sunshine
2016-06-20 10:09 ` Johannes Schindelin
2016-06-20 16:03 ` Junio C Hamano
2016-06-21 7:15 ` Johannes Schindelin
2016-06-21 16:50 ` Junio C Hamano
2016-06-22 7:24 ` Johannes Schindelin
2016-06-22 15:49 ` Junio C Hamano
2016-06-22 16:14 ` Johannes Schindelin
2016-06-22 17:37 ` Junio C Hamano
2016-06-22 17:53 ` Junio C Hamano
2016-06-20 10:55 ` [PATCH v2 0/7] Let log-tree and friends respect diffopt's `file` field Johannes Schindelin
2016-06-20 10:55 ` [PATCH v2 1/7] log-tree: respect diffopt's configured output file stream Johannes Schindelin
2016-06-20 17:01 ` Junio C Hamano
2016-06-21 7:31 ` Johannes Schindelin [this message]
2016-06-21 7:38 ` Johannes Schindelin
2016-06-21 10:39 ` Johannes Schindelin
2016-06-20 10:55 ` [PATCH v2 2/7] line-log: " Johannes Schindelin
2016-06-20 10:55 ` [PATCH v2 3/7] graph: respect the diffopt.file setting Johannes Schindelin
2016-06-20 10:55 ` [PATCH v2 4/7] shortlog: support outputting to streams other than stdout Johannes Schindelin
2016-06-20 10:55 ` [PATCH v2 5/7] format-patch: explicitly switch off color when writing to files Johannes Schindelin
2016-06-20 10:55 ` [PATCH v2 6/7] format-patch: avoid freopen() Johannes Schindelin
2016-06-20 10:56 ` [PATCH v2 7/7] format-patch: use stdout directly Johannes Schindelin
2016-06-20 18:57 ` Junio C Hamano
2016-06-20 11:50 ` [PATCH v2 0/7] Let log-tree and friends respect diffopt's `file` field Johannes Schindelin
2016-06-21 10:34 ` [PATCH v3 0/9] " Johannes Schindelin
2016-06-21 10:34 ` [PATCH v3 1/9] am: stop ignoring errors reported by log_tree_diff() Johannes Schindelin
2016-06-21 18:59 ` Junio C Hamano
2016-06-22 12:21 ` Johannes Schindelin
2016-06-21 10:34 ` [PATCH v3 2/9] Disallow diffopt.close_file when using the log_tree machinery Johannes Schindelin
2016-06-21 18:14 ` Junio C Hamano
2016-06-21 19:05 ` Junio C Hamano
2016-06-21 19:32 ` Junio C Hamano
2016-06-22 15:17 ` Johannes Schindelin
2016-06-21 10:34 ` [PATCH v3 3/9] log-tree: respect diffopt's configured output file stream Johannes Schindelin
2016-06-21 10:35 ` [PATCH v3 4/9] line-log: " Johannes Schindelin
2016-06-21 10:35 ` [PATCH v3 5/9] graph: respect the diffopt.file setting Johannes Schindelin
2016-06-21 10:35 ` [PATCH v3 6/9] shortlog: support outputting to streams other than stdout Johannes Schindelin
2016-06-21 10:35 ` [PATCH v3 7/9] format-patch: explicitly switch off color when writing to files Johannes Schindelin
2016-06-21 10:35 ` [PATCH v3 8/9] format-patch: avoid freopen() Johannes Schindelin
2016-06-21 10:35 ` [PATCH v3 9/9] format-patch: use stdout directly Johannes Schindelin
2016-06-21 13:47 ` [PATCH v3 0/9] Let log-tree and friends respect diffopt's `file` field Paul Tan
2016-06-21 14:12 ` Johannes Schindelin
2016-06-22 9:23 ` Paul Tan
2016-06-22 15:01 ` [PATCH v4 00/10] " Johannes Schindelin
2016-06-22 15:01 ` [PATCH v4 01/10] Prepare log/log-tree to reuse the diffopt.close_file attribute Johannes Schindelin
2016-06-24 20:56 ` Junio C Hamano
2016-06-26 6:56 ` Johannes Schindelin
2016-06-22 15:01 ` [PATCH v4 02/10] log-tree: respect diffopt's configured output file stream Johannes Schindelin
2016-06-22 15:01 ` [PATCH v4 03/10] line-log: " Johannes Schindelin
2016-06-22 15:01 ` [PATCH v4 04/10] graph: respect the diffopt.file setting Johannes Schindelin
2016-06-22 15:01 ` [PATCH v4 05/10] shortlog: support outputting to streams other than stdout Johannes Schindelin
2016-06-22 15:01 ` [PATCH v4 06/10] format-patch: explicitly switch off color when writing to files Johannes Schindelin
2016-06-24 22:01 ` Junio C Hamano
2016-06-26 6:49 ` Johannes Schindelin
2016-06-22 15:01 ` [PATCH v4 07/10] format-patch: avoid freopen() Johannes Schindelin
2016-06-22 15:02 ` [PATCH v4 08/10] format-patch: use stdout directly Johannes Schindelin
2016-06-24 22:03 ` Junio C Hamano
2016-06-22 15:02 ` [PATCH v4 09/10] shortlog: respect the --output=<file> setting Johannes Schindelin
2016-06-22 15:02 ` [PATCH v4 10/10] Ensure that log respects --output=<file> Johannes Schindelin
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
List information: http://vger.kernel.org/majordomo-info.html
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=alpine.DEB.2.20.1606210920510.22630@virtualbox \
--to=johannes.schindelin@gmx.de \
--cc=git@vger.kernel.org \
--cc=gitster@pobox.com \
--cc=sunshine@sunshineco.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
Code repositories for project(s) associated with this public inbox
https://80x24.org/mirrors/git.git
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).