bug-glibc
[Top][All Lists]
Advanced

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

Re: glibc 2.2.4: `make clean all' bails out due to a dependency problem


From: Maciej W. Rozycki
Subject: Re: glibc 2.2.4: `make clean all' bails out due to a dependency problem
Date: Thu, 23 Aug 2001 22:39:48 +0200 (MET DST)

On 23 Aug 2001, Ulrich Drepper wrote:

> >  I've been using `make clean all' since 2.0.94 up to 2.2.3 successfully. 
> > Such an invocation has a well-defined semantics -- see `info make'.
> 
> Just because you say so, right?  This is a completely unnecessary
> constraint on the Makefiles.

 I'm unsure what you mean, sorry.  What is the constraint you refer to? 
Do you mean multiple goals on a make's command line are the constraint? 
Please elaborate.

 Right now "$(common-objpfx)abi-versions.h" may be chosen by make to be
rebuilt at any time, even if the lack of "$(common-objpfx)sysd-sorted"
would prevent "$(common-objpfx)Versions.all" from being built.  The fact
make doesn't fail for `make all' (or simply `make') is pure coincidence.
The fact is there is a window when the rule for building
"$(common-objpfx)Versions.all" is unavailable and the one for
"$(common-objpfx)abi-versions.h" is still enabled.  My solution is to
disable the latter one during the window as well.  As a result the build
succeeds -- if anything else depended on "$(common-objpfx)abi-versions.h" 
during the window, the build would fail. 

 Of course there is another possibility -- maybe the rule for
"$(common-objpfx)Versions.all" shouldn't be disabled in the first place
and its invocation should just trigger making
"$(common-objpfx)sysd-sorted".  I can't say now -- I may investigate the
option if there is any negative impact due to my fix.  It has yet to be
observed, though -- I haven't been able to see any problems yet.

 The fact I use `make clean all' is purely a way to trigger the bug.  I
don't insist it's the best way to build glibc, nor encourage others to use
it.  It just shows there is an inconsistency in Makefiles and sooner or
later another rule may trigger it.  I may use your preferred way to build
glibc, what's the deal -- it won't fix the problem, though, only cover it
until someone else gets bitten by it somehow.  That would be unfortunate. 

  Maciej

-- 
+  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
+--------------------------------------------------------------+
+        e-mail: address@hidden, PGP key available        +




reply via email to

[Prev in Thread] Current Thread [Next in Thread]