Re: stacktrace vs. StackTrace

[Geoffrey Keating <geoffk at apple dot com>]

On 07/07/2006, at 11:58 AM, Michael Koch wrote:

> On Fri, Jul 07, 2006 at 11:48:36AM -0700, Geoffrey Keating wrote:
> > A recent checkin,
> >
> > --------------------------------------------------------------------- 
> > ---
> > r115235 | bryce | 2006-07-06 16:26:24 -0700 (Thu, 06 Jul 2006) | 9  
> > lines
> >
> > 2006-07-06  Bryce McKinlay <mckinlay@redhat.com>
> >
> >         * stacktrace.cc (ClassForFrame): Remove commented-out code.
> >         (UnwindTraceFn): Use _Unwind_GetIPInfo and adjust IP
> >         only when needed.
> >         (getLineNumberForFrame): Don't adjust IP here.
> >         * testsuite/libjava.lang/StackTrace.java: New test case.
> >         * testsuite/libjava.lang/StackTrace.out: Ditto.
> >
> >
> > added new files 'StackTrace.java' and 'StackTrace.out' even though
> > there was already 'stacktrace.java' and 'stacktrace.out'.
> >
> > On a case-insensitive filesystem, this causes 'svn update' to print:
> >
> > svn: Failed to add file 'StackTrace.out': object of the same name
> > already exists
> >
> > and the checkout fails.
> >
> > In order to fix this, I'm about to run the following commands:
> >
> > svn mv svn+ssh://gcc.gnu.org/svn/gcc/trunk/libjava/testsuite/
> > libjava.lang/{StackTrace,StackTrace2}.out
> > svn mv svn+ssh://gcc.gnu.org/svn/gcc/trunk/libjava/testsuite/
> > libjava.lang/{StackTrace,StackTrace2}.java
> >
> > and then probably I'll need to edit StackTrace2.java and/or
> > StackTrace2.out; I'll send a followup e-mail with that patch.
> >
> > I can't test this in advance because that would involve having a
> > checked-out tree, so there may be some temporary breakage.
> > Fortunately it should only affect the testsuite.
>
> Waht about "svn copy ...", editing the files, testing, "svn  
> remove ..."?

The tricky part there is "testing".  Without a self-consistent  
checkout, it would surely fail.

Attachment: smime.p7s
Description: S/MIME cryptographic signature