From: Jeff King <peff@peff.net>
To: Junio C Hamano <gitster@pobox.com>
Cc: Stefan Beller <sbeller@google.com>,
git@vger.kernel.org, jrnieder@gmail.com,
johannes.schindelin@gmail.com, Jens.Lehmann@web.de,
vlovich@gmail.com
Subject: Re: [PATCH 01/10] strbuf: Add strbuf_read_noblock
Date: Thu, 17 Sep 2015 14:02:06 -0400 [thread overview]
Message-ID: <20150917180205.GA29985@sigill.intra.peff.net> (raw)
In-Reply-To: <xmqqoah1hqxs.fsf@gitster.mtv.corp.google.com>
On Thu, Sep 17, 2015 at 10:54:39AM -0700, Junio C Hamano wrote:
> OK. Trying to repurpose strbuf_read() for non-blocking FD was a
> silly idea to begin with, as it wants to read thru to the EOF, and
> setting FD explicitly to O_NONBLOCK is a sign that the caller wants
> to grab as much as possible and does not want to wait for the EOF.
>
> So assuming we want strbuf_read_nonblock(), what interface do we
> want from it? We could:
>
> * Have it grab as much as possible into sb as long as it does not
> block?
>
> * Have it grab reasonably large amount into sb, and not blocking is
> an absolute requirement, but the function is not required to read
> everything that is available on the FD (i.e. the caller is
> expected to loop)?
I think we are crossing emails, but I would definitely argue for the
latter.
> If we choose the latter, then your "EAGAIN? EOF?" becomes easier,
> no? We only have to do a single call to xread(), and then we:
>
> - get EAGAIN or EWOULDBLOCK; leave sb as-is, set errno==EAGAIN and
> return -1.
>
> - get something (in which case that is not an EOF yet); append to
> sb, return the number of bytes.
>
> - get EOF; leave sb as-is, return 0.
Yes, exactly.
> ssize_t strbuf_read_nonblock(struct strbuf *sb, int fd, size_t hint)
> {
> strbuf_grow(sb, hint ? hint : 8192);
> ssize_t want = sb->alloc - sb->len - 1;
> ssize_t got = xread_nonblock(fd, sb->buf + sb->len, want);
> if (got < 0) {
> if (errno == EWOULDBLOCK)
> errno = EAGAIN; /* make life easier for the caller */
> return got;
> }
I like the normalizing of errno, but that should probably be part of
xread_nonblock(), I would think.
> sb->len += got;
> sb->buf[sb->len] = '\0';
Use strbuf_setlen() here?
> return got;
If "got == 0", we naturally do not change sb->len at all, and the strbuf
is left unchanged. But do we want to de-allocate what we allocated in
strbuf_grow() above? That is what strbuf_read() does, but I think it is
even less likely to help anybody here. With the original strbuf_read(),
you might do:
if (strbuf_read(&buf, fd, hint) <= 0)
return; /* got nothing */
but because the nature of strbuf_read_nonblock() is to call it from a
loop, you'd want to strbuf_release() when you leave the loop anyway.
-Peff
next prev parent reply other threads:[~2015-09-17 18:02 UTC|newest]
Thread overview: 37+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-09-17 1:38 [PATCH 00/10] fetch submodules in parallel and a preview on parallel "submodule update" Stefan Beller
2015-09-17 1:38 ` [PATCH 01/10] strbuf: Add strbuf_read_noblock Stefan Beller
2015-09-17 16:13 ` Junio C Hamano
2015-09-17 16:30 ` Jeff King
2015-09-17 16:44 ` Junio C Hamano
2015-09-17 16:51 ` Stefan Beller
2015-09-17 16:57 ` Jeff King
2015-09-17 16:58 ` Junio C Hamano
2015-09-17 17:13 ` Jeff King
2015-09-17 17:26 ` Stefan Beller
2015-09-17 17:35 ` Jeff King
2015-09-17 17:45 ` Stefan Beller
2015-09-17 17:50 ` Jeff King
2015-09-17 17:53 ` Stefan Beller
2015-09-17 17:57 ` Junio C Hamano
2015-09-17 17:54 ` Junio C Hamano
2015-09-17 18:02 ` Jeff King [this message]
2015-09-17 17:20 ` Stefan Beller
2015-09-17 1:39 ` [PATCH 02/10] run-command: factor out return value computation Stefan Beller
2015-09-17 10:33 ` Jeff King
2015-09-17 1:39 ` [PATCH 03/10] run-command: add an asynchronous parallel child processor Stefan Beller
2015-09-17 21:44 ` Junio C Hamano
2015-09-17 23:19 ` Stefan Beller
2015-09-18 1:05 ` Junio C Hamano
2015-09-18 16:36 ` Stefan Beller
2015-09-17 1:39 ` [PATCH 04/10] fetch_populated_submodules: use new parallel job processing Stefan Beller
2015-09-17 1:39 ` [PATCH 05/10] submodules: Allow parallel fetching, add tests and documentation Stefan Beller
2015-09-17 1:39 ` [PATCH 06/10] git submodule update: Redirect any output to stderr Stefan Beller
2015-09-17 20:31 ` Eric Sunshine
2015-09-17 20:38 ` Stefan Beller
2015-09-17 1:39 ` [PATCH 07/10] git submodule update: pass --prefix only with a non empty prefix Stefan Beller
2015-09-17 20:33 ` Eric Sunshine
2015-09-17 1:39 ` [PATCH 08/10] git submodule update: cmd_update_recursive Stefan Beller
2015-09-17 1:39 ` [PATCH 09/10] " Stefan Beller
2015-09-17 20:37 ` Eric Sunshine
2015-09-17 1:39 ` [PATCH 10/10] git submodule update: cmd_update_fetch Stefan Beller
2015-09-17 17:06 ` [PATCH 00/10] fetch submodules in parallel and a preview on parallel "submodule update" Jacob Keller
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=20150917180205.GA29985@sigill.intra.peff.net \
--to=peff@peff.net \
--cc=Jens.Lehmann@web.de \
--cc=git@vger.kernel.org \
--cc=gitster@pobox.com \
--cc=johannes.schindelin@gmail.com \
--cc=jrnieder@gmail.com \
--cc=sbeller@google.com \
--cc=vlovich@gmail.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).