public inbox for gcc-prs@sourceware.org
help / color / mirror / Atom feed
From: benko@sztaki.hu
To: gcc-gnats@gcc.gnu.org
Subject: libstdc++/6047: weird behaviour when reading from a pipe through an ifstream
Date: Sat, 23 Mar 2002 01:16:00 -0000	[thread overview]
Message-ID: <20020323090914.28476.qmail@sources.redhat.com> (raw)


>Number:         6047
>Category:       libstdc++
>Synopsis:       weird behaviour when reading from a pipe through an ifstream
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sat Mar 23 01:16:00 PST 2002
>Closed-Date:
>Last-Modified:
>Originator:     Benko Pal
>Release:        3.0.4
>Organization:
>Environment:
i686, SuSE 7.3, glibc 2.2.4
>Description:
When two programs communicate through fifos and handles them
with std::fstreams, then flushing on the writing side does
not make the contents appear on the reading side.  The
reading side only sees the data when the writing buffer is
overrun or the writing side closes the buffer.

A test case:

------------------------------------------
#ifdef WORKAROUND
# include <cstdio>
#endif /* WORKAROUND */
#include <fstream>
#include <iostream>


int
main()
{
#ifdef WORKAROUND
  std::filebuf b(std::fopen("/tmp/pipe", "r"), std::ios_base::in, 1);
  std::istream in(&b);
#else /* ! WORKAROUND */
  std::ifstream in("/tmp/pipe");
#endif /* WORKAROUND */
  char          c;

  while (in.get(c))
    std::cout.put(c);
  
  return 0;
}
-------------------------------------------

In an other shell do

$ rm -f /tmp/pipe 
$ mkfifo /tmp/pipe
$ cat > /tmp/pipe
hallo
world
^D
$ 

If the program above is compiled with 3.0.*, and run, then
"hallo" and "world" appear together, after ^D is pressed in
the writing shell.
Compiling with 2.95.3, "hallo" appears when I hit <enter> in
the writing shell, separately from "world", so this is a
sort of regression.  An equivalent C program behaves like
2.95.3, compiled with either compiler.  This behavior is
very important for me, without this the two programs just
keep waiting for each other.

By defining WORKAROUND I get the expected result, but it's a
bit inconvenient to initialise a stream like this.
>How-To-Repeat:

>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted:


             reply	other threads:[~2002-03-23  9:16 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2002-03-23  1:16 benko [this message]
2002-04-09 12:20 bkoz
2002-11-06 13:53 bangerth
2002-11-30 20:06 paolo

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

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20020323090914.28476.qmail@sources.redhat.com \
    --to=benko@sztaki.hu \
    --cc=gcc-gnats@gcc.gnu.org \
    /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.
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).