Re: Patch for gdb/mi 792

[Andrew Cagney <ac131313 at redhat dot com>]

> On Tue, 22 Oct 2002 18:12:10 -0400, Andrew Cagney <ac131313@redhat.com> said:
>
>
> > Oh, and I suspect you've been hit.  It's detected some more C++
> > testsuite filename problems.
>
> No, it's the same one that I already fixed a while ago.  I'm not sure
> why the ARI isn't noticing that; maybe there's a bug in the script?
> It seems to me that there is the correct info in fnchange.lst.
The ARI script looks like:

    echo "`date`: Checking for doschk" 1>&2
    rm -f "${wwwdir}"/ari.doschk.*
    fnchange_lst="${srcdir}"/gdb/config/djgpp/fnchange.lst
    fnchange_sed="${wwwdir}"/ari.doschk.sed
    # Create fnchange.sed
    sed -e 's;@[^@]*@[/]*\([^ ]*\) @[^@]*@[/]*\([^ ]*\);s,\1,\2,;' \
        < "${fnchange_lst}" > "${fnchange_sed}"
    # Do the raw analysis
    ( cd "${srcdir}" && find * \
        -name '*.info-[0-9]*' -prune \
        -o -name tcl -prune \
        -o -name itcl -prune \
        -o -name tk -prune \
        -o -name libgui -prune \
        -o -name tix -prune \
        -o -name dejagnu -prune \
        -o -name expect -prune \
        -o -type f -print ) | sed -f ${fnchange_sed} \
    | doschk > ${wwwdir}/ari.doschk.out
    # Magic to map ari.doschk.out to ari.doschk.bug goes here

A quick poke shows it's trying to apply the two transformations:

s,gdb/testsuite/gdb.c++,gdb/testsuite/gdb.cxx,
...
s,gdb/testsuite/gdb.c++/m-static.cc,gdb/testsuite/gdb.cxx/m-static.cc,
s,gdb/testsuite/gdb.c++/m-static1.cc,gdb/testsuite/gdb.cxx/m-stat1.cc,
s,gdb/testsuite/gdb.c++/m-static.exp,gdb/testsuite/gdb.cxx/m-static.exp,
s,gdb/testsuite/gdb.c++/m-static.h,gdb/testsuite/gdb.cxx/m-static.h,

but the actual renames never match since the first s/// resulted in:

gdb/testsuite/gdb.cxx/m-static.cc
gdb/testsuite/gdb.cxx/m-static.exp
gdb/testsuite/gdb.cxx/m-static.h
gdb/testsuite/gdb.cxx/m-static1.cc

being passed to doschk (see src/contrib).

I'm not sure how well this compares to the behavior of the actual DJGPP 
command.

Andrew