[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#52373: Wishlist bug?



At Sun, 5 Jan 2003 18:06:08 +0000,
Colin Watson wrote:
> On Sun, Dec 29, 2002 at 01:32:17AM +0900, GOTO Masanori wrote:
> > At Thu, 26 Dec 2002 22:40:14 -0500,
> > H. S. Teoh <hsteoh@quickfur.ath.cx> wrote:
> > > Should this bug be downgraded to wishlist? The bug submitter did say "it
> > > would be useful if ...". Moreover, according to glibc info docs:
> > > 
> > >  - Macro: int EPIPE
> > >      Broken pipe; there is no process reading from the other end of a
> > >      pipe.  Every library function that returns this error code also
> > >      generates a `SIGPIPE' signal; this signal terminates the program
> > >      if not handled or blocked.  Thus, your program will never actually
> > >      see `EPIPE' unless it has handled or blocked `SIGPIPE'.
> > > 
> > > So the two seems to be quite related to each other, and I don't see any
> > > compelling reason to distinguish between error messages generated by
> > > either one. Sounds very much like wishlist to me.
> > 
> > I agree.
> > 
> > SIGPIPE is one of signals, and EPIPE is one of error numbers.  The
> > message "Broken pipe" is the result of SIGPIPE, and its error number.
> > I think there is no merit to distinguish them.  Moreover, how to
> > change the message between the two? It's difficult.
> 
> For what it's worth, I remember Owen talking about this problem back in
> 1999. He was trying to develop a "guru" locale which simply printed the
> error or signal name - that is, "EPIPE" or "SIGPIPE" - rather than an
> explanatory message. This can't be done correctly when the error
> messages for EPIPE and SIGPIPE are the same, because the gettext msgid
> is keyed on the description, not the error or signal name.

"guru" message seems good idea, but is it _really_ needed?
Is it a real requirement?

> > So I think it's not bug, it's not wishlist item. I close this bug.
> 
> I think it's a valid wishlist item and should be reopened as such, even
> if it may be difficult to fix.

I think this patch fixes this wishlist, but I have not tested yet.


--- sysdeps/unix/sysv/sysv4/solaris2/sparc/errlist.c	2001-07-06 13:56:22.000000000 +0900
+++ sysdeps/unix/sysv/sysv4/solaris2/sparc/errlist.c.new	2003-01-06 22:37:05.000000000 +0900
@@ -59,7 +59,7 @@
   N_("Illegal seek"),
   N_("Read-only file system"),
   N_("Too many links"),
-  N_("Broken pipe"),
+  N_("Broken pipe error"),
   N_("Argument out of domain"),
   N_("Result too large"),
   N_("No message of desired type"),
--- sysdeps/gnu/errlist.c	2002-08-29 08:53:03.000000000 +0900
+++ sysdeps/gnu/errlist.c.new	2003-01-06 22:37:56.000000000 +0900
@@ -348,7 +348,7 @@
 TRANS @code{SIGPIPE} signal; this signal terminates the program if not handled
 TRANS or blocked.  Thus, your program will never actually see @code{EPIPE}
 TRANS unless it has handled or blocked @code{SIGPIPE}. */
-    [ERR_REMAP (EPIPE)] = N_("Broken pipe"),
+    [ERR_REMAP (EPIPE)] = N_("Broken pipe error"),
 # if EPIPE > ERR_MAX
 # undef ERR_MAX
 # define ERR_MAX EPIPE



Reply to: