[PATCH] Fix spurious mi-basics.exp failures

[PATCH] Fix spurious mi-basics.exp failures

[Ulrich Weigand]

Hello,

I'm seeing some FAILs in the mi-basics test case.  They're apparently
caused by the expected output regexps in some of the mi_gdb_test
statements lacking the 'serial number' the MI output lines start with.

Most mi_gdb_test statments already contain that number, but the 202,
203, and 208 tests don't have it.  Those are exactly the tests that
fail for me.  Adding the numbers fixes the test.

Tested on s390-ibm-linux and s390x-ibm-linux.
OK for mainline?


ChangeLog:

	* gdb.mi/mi-basics.exp: Add missing serial number to 
	expected output in some mi_gdb_test statements.
	* gdb.mi/mi2-basics.exp: Likewise.

Index: gdb/testsuite/gdb.mi/mi-basics.exp
===================================================================
RCS file: /cvs/src/src/gdb/testsuite/gdb.mi/mi-basics.exp,v
retrieving revision 1.12
diff -c -p -r1.12 mi-basics.exp
*** gdb/testsuite/gdb.mi/mi-basics.exp	27 Sep 2005 22:39:01 -0000	1.12
--- gdb/testsuite/gdb.mi/mi-basics.exp	6 May 2006 00:55:38 -0000
*************** proc test_dir_specification {} {
*** 160,170 ****
      set envirodir [string_to_regexp ${srcdir}/${subdir}]
  
      mi_gdb_test "202-environment-directory ${srcdir}/${subdir}" \
!              "\\\^done,source-path=\"${envirodir}.\\\$cdir.\\\$cwd\"" \
               "environment-directory arg operation"
  
      mi_gdb_test "203-environment-directory" \
!              "\\\^done,source-path=\"${envirodir}.\\\$cdir.\\\$cwd\"" \
               "environment-directory empty-string operation"
  
      mi_gdb_test "204-environment-directory -r" \
--- 160,170 ----
      set envirodir [string_to_regexp ${srcdir}/${subdir}]
  
      mi_gdb_test "202-environment-directory ${srcdir}/${subdir}" \
!              "202\\\^done,source-path=\"${envirodir}.\\\$cdir.\\\$cwd\"" \
               "environment-directory arg operation"
  
      mi_gdb_test "203-environment-directory" \
!              "203\\\^done,source-path=\"${envirodir}.\\\$cdir.\\\$cwd\"" \
               "environment-directory empty-string operation"
  
      mi_gdb_test "204-environment-directory -r" \
*************** proc test_path_specification {} {
*** 223,229 ****
               "environment-path no-args operation"
  
      mi_gdb_test "208-environment-path $srcdir $objdir" \
!              "\\\^done,path=\"$escapedsrcdir.$escapedobjdir.$orig_path\"" \
               "environment-path dir1 dir2 operation"
  
      mi_gdb_test "209-environment-path -r $objdir" \
--- 223,229 ----
               "environment-path no-args operation"
  
      mi_gdb_test "208-environment-path $srcdir $objdir" \
!              "208\\\^done,path=\"$escapedsrcdir.$escapedobjdir.$orig_path\"" \
               "environment-path dir1 dir2 operation"
  
      mi_gdb_test "209-environment-path -r $objdir" \
Index: gdb/testsuite/gdb.mi/mi2-basics.exp
===================================================================
RCS file: /cvs/src/src/gdb/testsuite/gdb.mi/mi2-basics.exp,v
retrieving revision 1.4
diff -c -p -r1.4 mi2-basics.exp
*** gdb/testsuite/gdb.mi/mi2-basics.exp	27 Sep 2005 22:39:03 -0000	1.4
--- gdb/testsuite/gdb.mi/mi2-basics.exp	6 May 2006 00:55:38 -0000
*************** proc test_dir_specification {} {
*** 161,171 ****
      set envirodir [string_to_regexp ${srcdir}/${subdir}]
  
      mi_gdb_test "202-environment-directory ${srcdir}/${subdir}" \
!              "\\\^done,source-path=\"${envirodir}.\\\$cdir.\\\$cwd\"" \
               "environment-directory arg operation"
  
      mi_gdb_test "203-environment-directory" \
!              "\\\^done,source-path=\"${envirodir}.\\\$cdir.\\\$cwd\"" \
               "environment-directory empty-string operation"
  
      mi_gdb_test "204-environment-directory -r" \
--- 161,171 ----
      set envirodir [string_to_regexp ${srcdir}/${subdir}]
  
      mi_gdb_test "202-environment-directory ${srcdir}/${subdir}" \
!              "202\\\^done,source-path=\"${envirodir}.\\\$cdir.\\\$cwd\"" \
               "environment-directory arg operation"
  
      mi_gdb_test "203-environment-directory" \
!              "203\\\^done,source-path=\"${envirodir}.\\\$cdir.\\\$cwd\"" \
               "environment-directory empty-string operation"
  
      mi_gdb_test "204-environment-directory -r" \
*************** proc test_path_specification {} {
*** 226,232 ****
               "environment-path no-args operation"
  
      mi_gdb_test "208-environment-path $srcdir $objdir" \
!              "\\\^done,path=\"$escapedsrcdir.$escapedobjdir.$orig_path\"" \
               "environment-path dir1 dir2 operation"
  
      mi_gdb_test "209-environment-path -r $objdir" \
--- 226,232 ----
               "environment-path no-args operation"
  
      mi_gdb_test "208-environment-path $srcdir $objdir" \
!              "208\\\^done,path=\"$escapedsrcdir.$escapedobjdir.$orig_path\"" \
               "environment-path dir1 dir2 operation"
  
      mi_gdb_test "209-environment-path -r $objdir" \
-- 
  Dr. Ulrich Weigand
  Linux on zSeries Development
  Ulrich.Weigand@de.ibm.com

Re: [PATCH] Fix spurious mi-basics.exp failures

[Bob Rossi]

On Sat, May 06, 2006 at 03:19:32AM +0200, Ulrich Weigand wrote:
> Hello,
> 
> I'm seeing some FAILs in the mi-basics test case.  They're apparently
> caused by the expected output regexps in some of the mi_gdb_test
> statements lacking the 'serial number' the MI output lines start with.
> 
> Most mi_gdb_test statments already contain that number, but the 202,
> 203, and 208 tests don't have it.  Those are exactly the tests that
> fail for me.  Adding the numbers fixes the test.

hmm, I would be surprised if that was the only reason those tests are
failing. Are you sure there isn't more to it? Are the commands
implemented yet properly?

Bob Rossi

Re: [PATCH] Fix spurious mi-basics.exp failures

[Daniel Jacobowitz]

On Sat, May 06, 2006 at 03:19:32AM +0200, Ulrich Weigand wrote:
> Hello,
> 
> I'm seeing some FAILs in the mi-basics test case.  They're apparently
> caused by the expected output regexps in some of the mi_gdb_test
> statements lacking the 'serial number' the MI output lines start with.
> 
> Most mi_gdb_test statments already contain that number, but the 202,
> 203, and 208 tests don't have it.  Those are exactly the tests that
> fail for me.  Adding the numbers fixes the test.
> 
> Tested on s390-ibm-linux and s390x-ibm-linux.
> OK for mainline?
> 
> 
> ChangeLog:
> 
> 	* gdb.mi/mi-basics.exp: Add missing serial number to 
> 	expected output in some mi_gdb_test statements.
> 	* gdb.mi/mi2-basics.exp: Likewise.

Are you sure you tested these in the same build tree where they were
broken?

I might be misremembering, but when these tests were added, I think I
complained about the regexps: they fail when the srcdir is specified
using a relative path, but succeed with an absolute path, or the other
way round.

203-environment-directory
203^done,source-path="/big/fsf/commit/src/gdb/testsuite/gdb.mi:$cdir:$cwd"
(gdb) 
PASS: gdb.mi/mi-basics.exp: environment-directory empty-string
operation

Then again, my log does seem to match your results... I don't remember
these failing on this machine before.

From my previous system, an old test run passed with the anchors:

202-environment-directory /big/fsf/commit/src/gdb/testsuite/gdb.mi
202^done,source-path="/big/fsf/commit/src/gdb/testsuite/gdb.mi:$cdir:$cwd"
(gdb) 
PASS: gdb.mi/mi-basics.exp: environment-directory arg operation
203-environment-directory
203^done,source-path="/big/fsf/commit/src/gdb/testsuite/gdb.mi:$cdir:$cwd"
(gdb) 
PASS: gdb.mi/mi-basics.exp: environment-directory empty-string operation


From today:
202-environment-directory /space/fsf/commit/src/gdb/testsuite/gdb.mi
202^done,source-path="/space/fsf/commit/src/gdb/testsuite/gdb.mi:$cdir:$cwd"
(gdb)
FAIL: gdb.mi/mi-basics.exp: environment-directory arg operation
203-environment-directory
203^done,source-path="/space/fsf/commit/src/gdb/testsuite/gdb.mi:$cdir:$cwd"
(gdb)
FAIL: gdb.mi/mi-basics.exp: environment-directory empty-string operation


Oh, I bet I know what's happened!  These are the three tests that
failed based on relative srcdirs or something like that.  So when Bob
committed fixes to the MI testsuite which added anchoring to
mi_gdb_test, they probably failed for him both before and after, and
that's how they were missed.

Bob, does that sound reasonable?  If so, the patch is OK, and thanks!


-- 
Daniel Jacobowitz
CodeSourcery

Re: [PATCH] Fix spurious mi-basics.exp failures

[Bob Rossi]

On Fri, May 05, 2006 at 09:52:10PM -0400, Daniel Jacobowitz wrote:
> On Sat, May 06, 2006 at 03:19:32AM +0200, Ulrich Weigand wrote:
> > Hello,
> > 
> > I'm seeing some FAILs in the mi-basics test case.  They're apparently
> > caused by the expected output regexps in some of the mi_gdb_test
> > statements lacking the 'serial number' the MI output lines start with.
> > 
> > Most mi_gdb_test statments already contain that number, but the 202,
> > 203, and 208 tests don't have it.  Those are exactly the tests that
> > fail for me.  Adding the numbers fixes the test.
> > 
> > Tested on s390-ibm-linux and s390x-ibm-linux.
> > OK for mainline?
> > 
> > 
> > ChangeLog:
> > 
> > 	* gdb.mi/mi-basics.exp: Add missing serial number to 
> > 	expected output in some mi_gdb_test statements.
> > 	* gdb.mi/mi2-basics.exp: Likewise.
> 
> Are you sure you tested these in the same build tree where they were
> broken?
> 
> I might be misremembering, but when these tests were added, I think I
> complained about the regexps: they fail when the srcdir is specified
> using a relative path, but succeed with an absolute path, or the other
> way round.
> 
> 203-environment-directory
> 203^done,source-path="/big/fsf/commit/src/gdb/testsuite/gdb.mi:$cdir:$cwd"
> (gdb) 
> PASS: gdb.mi/mi-basics.exp: environment-directory empty-string
> operation
> 
> Then again, my log does seem to match your results... I don't remember
> these failing on this machine before.
> 
> >From my previous system, an old test run passed with the anchors:
> 
> 202-environment-directory /big/fsf/commit/src/gdb/testsuite/gdb.mi
> 202^done,source-path="/big/fsf/commit/src/gdb/testsuite/gdb.mi:$cdir:$cwd"
> (gdb) 
> PASS: gdb.mi/mi-basics.exp: environment-directory arg operation
> 203-environment-directory
> 203^done,source-path="/big/fsf/commit/src/gdb/testsuite/gdb.mi:$cdir:$cwd"
> (gdb) 
> PASS: gdb.mi/mi-basics.exp: environment-directory empty-string operation
> 
> 
> >From today:
> 202-environment-directory /space/fsf/commit/src/gdb/testsuite/gdb.mi
> 202^done,source-path="/space/fsf/commit/src/gdb/testsuite/gdb.mi:$cdir:$cwd"
> (gdb)
> FAIL: gdb.mi/mi-basics.exp: environment-directory arg operation
> 203-environment-directory
> 203^done,source-path="/space/fsf/commit/src/gdb/testsuite/gdb.mi:$cdir:$cwd"
> (gdb)
> FAIL: gdb.mi/mi-basics.exp: environment-directory empty-string operation
> 
> 
> Oh, I bet I know what's happened!  These are the three tests that
> failed based on relative srcdirs or something like that.  So when Bob
> committed fixes to the MI testsuite which added anchoring to
> mi_gdb_test, they probably failed for him both before and after, and
> that's how they were missed.
> 
> Bob, does that sound reasonable?  If so, the patch is OK, and thanks!

I can't think of exactly how the anchoring would have just anything
related to these tests.  I'm just wondering, what fixed the 
relative/absolute path problem?

This is what I got from what I just ran,
202-environment-directory ../../../src/gdb/testsuite/gdb.mi^M 202^done,source-path="/home/bob/cvs/gdb/original/builddir/gdb/testsuite/../../../src/gdb/testsuite/gdb.mi:$cdir:$cwd"^M
(gdb) ^M
FAIL: environment-directory arg operation
203-environment-directory^M
203^done,source-path="/home/bob/cvs/gdb/original/builddir/gdb/testsuite/../../../src/gdb/testsuite/gdb.mi:$cdir:$cwd"^M
(gdb) ^M
FAIL: environment-directory empty-string operation

Bob Rossi

Re: [PATCH] Fix spurious mi-basics.exp failures

[Daniel Jacobowitz]

On Fri, May 05, 2006 at 10:02:37PM -0400, Bob Rossi wrote:
> I can't think of exactly how the anchoring would have just anything
> related to these tests.  I'm just wondering, what fixed the 
> relative/absolute path problem?

My point is that before the anchoring, the "^done" pattern probably
matched "203^done" in the output.

They're still broken if you configure with a relative path, which you
can see from your output:

> 202-environment-directory ../../../src/gdb/testsuite/gdb.mi^M 202^done,source-path="/home/bob/cvs/gdb/original/builddir/gdb/testsuite/../../../src/gdb/testsuite/gdb.mi:$cdir:$cwd"^M

> 203^done,source-path="/home/bob/cvs/gdb/original/builddir/gdb/testsuite/../../../src/gdb/testsuite/gdb.mi:$cdir:$cwd"^M

Mine doesn't look like that.

-- 
Daniel Jacobowitz
CodeSourcery

Re: [PATCH] Fix spurious mi-basics.exp failures

[Ulrich Weigand]

Daniel Jacobowitz wrote:

> On Fri, May 05, 2006 at 10:02:37PM -0400, Bob Rossi wrote:
> > I can't think of exactly how the anchoring would have just anything
> > related to these tests.  I'm just wondering, what fixed the 
> > relative/absolute path problem?
> 
> My point is that before the anchoring, the "^done" pattern probably
> matched "203^done" in the output.
> 
> They're still broken if you configure with a relative path, which you
> can see from your output:

That sounds like a reasonable explanation.  I've always had to configure
with an absolute path, or else these tests would fail.  However, since
some time the tests fail even when configuring with an absolute path.
The patch I'm proposing gets them back to passing in that case (with
relative path they still fail).

Is the patch OK then?

Bye,
Ulrich

-- 
  Dr. Ulrich Weigand
  Linux on zSeries Development
  Ulrich.Weigand@de.ibm.com

Re: [PATCH] Fix spurious mi-basics.exp failures

[Daniel Jacobowitz]

On Sat, May 06, 2006 at 06:57:33PM +0200, Ulrich Weigand wrote:
> Daniel Jacobowitz wrote:
> 
> > On Fri, May 05, 2006 at 10:02:37PM -0400, Bob Rossi wrote:
> > > I can't think of exactly how the anchoring would have just anything
> > > related to these tests.  I'm just wondering, what fixed the 
> > > relative/absolute path problem?
> > 
> > My point is that before the anchoring, the "^done" pattern probably
> > matched "203^done" in the output.
> > 
> > They're still broken if you configure with a relative path, which you
> > can see from your output:
> 
> That sounds like a reasonable explanation.  I've always had to configure
> with an absolute path, or else these tests would fail.  However, since
> some time the tests fail even when configuring with an absolute path.
> The patch I'm proposing gets them back to passing in that case (with
> relative path they still fail).
> 
> Is the patch OK then?

Yes, it is.  Thanks a lot.

-- 
Daniel Jacobowitz
CodeSourcery