Tester: Jon A. Christopher (JAC) <jac8792@tamu.edu>
	SGI-Motif 1.2.4
	SGI-LessTif 0.85.2 ++

test1: FAIL
test2: PASS (except as noted in test1)
test3: FAIL
test4: PASS (except as noted in test1)
test5: PASS (except as noted in test1)
test6: FAIL
test7: FAIL
test8: PASS (except as noted in test1)
test9: UNKNOWN
test10: UNKNOWN
test11: UNKNOWN

==================================================
test1: 
Jun 20 1998: JAC

Test appears to be designed to print the value of FileSB resources from a
callback. 

Ok, here's the rundown on pressing return in FileSB when various areas
have focus from SGI's motif 1.2.4:

Filter TextF: When this widget gets focus, the "Filter" button gets
surrounded by a etched shadow.  This is probably a SGI enhancement, but it
provides a visual cue as to what the default action now is.  Lesstif does
not have such a shadow, even if you press the buttons directly. Pressing
return activates the filter button.  Lesstif does this properly. 

Directories ScrolledList: When this widget gets focus, Filter is also
shadowed as described above.  Pressing return causes the directory to
change to the selected directory.  This is what lesstif does, too.

Files ScrolledList: "Ok" gets shadowed.  Pressing enter activates Ok with
the for the selected file. Lesstif does this too.

Selection TextF: Ok shadowed.  Return activates ok.  Lesstif doesn't do
this.

rws 20 Jun 1998 The pressing of Return not working is because we are 
missing the KeyReturn action, I think, in either SelectionBox or List.

Double click behavior: The lesstif version requires a much faster
double-click speed than the Motif version.  On a single click of a file in
the file list, both versions put the file name in the selection widget.
On a double click, both print out the list of resources, similar to the Ok
button press.  However, after a double click, the motif version replaces
the selection string with "test it out", while the lesstif version does
not.  The double click speed may be a problem with the scrolled list
widget, not FileSB directly.

There were also some minor geometry differences; the motif version was a
little wider.

No other differences were apparent in a casual inspection.
==================================================
test10:
un-tested

Karsten's test to print out the defaultButton all over the place.

==================================================
test11:
un-tested

Eric Howe <mu@clio.trends.ca>, of mgv fame, sent us this to show what happens
during initialize when the dirMask is specified but not the pattern or
directory.

==================================================
test2:
Jun 20 1998 JAC

This test appears to be a PushButton-activated version of test1.  See the
above comments.  Other than things noted above, this test appears to work
properly.
==================================================
test3:
Jun 20 1998 JAC

Test initially created to simulate a problem in nedit.
Then augmented to observe callback structures.

* test sequence 
#/bin/tcsh
./test3.motif >& motif.out
./test3 >& lesstif.out
diff lesstif.out motif.out

In each application:
Pushme->Ok->Pushme->Filter->Cancel->Help->Double click in Filter TextF->DC
on first entry in directories->DC on first entry in files->DC in Selection
TF.

Result:
 diff lesstif.out motif.out 
2,6c2,6
<       Reason XmCR_OK, event 7fff2b00
<       value '/usr/people/jon/src/lesstif/test/Xm/filesb/', length 60
<       mask '/usr/people/jon/src/lesstif/test/Xm/filesb/*', length 61
<       dir '/usr/people/jon/src/lesstif/test/Xm/filesb/', length 60
<       pattern '*', length 18
---
>       Reason XmCR_OK, event 7fff279c
>       value '/usr/people/jon/src/lesstif/test/Xm/filesb/', length 52
>       mask '/usr/people/jon/src/lesstif/test/Xm/filesb/*', length 53
>       dir '/usr/people/jon/src/lesstif/test/Xm/filesb/', length 52
>       pattern '*', length 10
9c9
<       Reason XmCR_OK, event 7fff2b00
---
>       Reason XmCR_OK, event 7fff2ac0
11,13c11,13
<       mask '/usr/people/jon/src/lesstif/test/Xm/filesb/*', length 61
<       dir '/usr/people/jon/src/lesstif/test/Xm/filesb/', length 60
<       pattern '*', length 18
---
>       mask '/usr/people/jon/src/lesstif/test/Xm/filesb/*', length 53
>       dir '/usr/people/jon/src/lesstif/test/Xm/filesb/', length 52
>       pattern '*', length 10

We can of course ignore the event numbers, but what about the lengths?
I'm calling this a failure for now.

==================================================
test4:
Jun 20 1998 JAC

Generated by X-designer: test of fileSearchProc, modify default to test
for writable files.

Lesstif and Motif work the same.

==================================================
test5:
Jun 20 1998 JAC

Hey, I wrote this test!  This test demonstrated a bug (since fixed) where
adding a MenuBar to a FileSB wasn't positioned properly under lesstif.

Lesstif and Motif work the same.

==================================================
test6:
Jun 20 1998 JAC

Same as test4, but with the file list unmanaged.

There are geometry differences:
The motif version is wide enough for the entire filter string, while the
Lesstif version is not.  In the motif version the string "Files" appears,
although the file list does not.  In the lesstif version neither the
string nor the file list appears.  

Looks like we forgot to tie the label management to the window.

==================================================
test7:
Jun 20 1998 JAC

I wrote this one too.  From the test header comments:

  I've noticed that FileSelectionDialogs do not handle geometry correctly
  under certain cases.  If there's a menubar in the dialog, but no managed
  buttons in the menu bar, any action (filter, double clicking a
  directory, etc) causes the dialog to grow larger.  SGI Motif 1.2 doesn't
  exhibit this behavior.

The worst of these problems have been fixed now, but there's still some
weirdness.

Motif: Press cancel, notice that Button in the menu bar disappears.  The
menu bar shrinks (but is still visible, I'd guess 20 pixels, top/bottom
borders, perhaps).  Press again, Button is restored.

Lesstif: Press Cancel, Button disappears, MenuBar does *not* shrink. Press
again, Button reappears, but bottom shadow or menubar disappears.  Press
again, shadow reappears, Button disappears.  Continued presses cycle
through behaviors now.

Pressing "Filter" when the Button is gone makes the menubar disappear, and
you have to play around with various combinations of filter and cancel to
get the button/menu bar to reappear.

Jun 24 1998: Update: Rick has partially fixed this.  Currently, under
lesstif, multiple cancel presses manage and unmanage the button cleanly.

Under Motif, when the button is unmanaged, the geometry is recalculated,
and the menu bar shrinks, but nothing is apparently done to the geometry
when filter is pressed.  Under lesstif, nothing is done to the geometry
when cancel is pressed, but pressing filter makes the whole menubar
disappear. 

==================================================
test8:
Jun 20 1998 JAC

Same as test4 with the resize policy set to RESIZE_GROW.  To properly
evaluate this test, you must compare test4 to test8.  Start in a directory
with fairly short file names, and then navagate into a directory with
really long file names.  The entire dialog box should grow to accomidate
the wide names.  In test4, if you then go back to a dir with shorter
names, the dialog will shrink, but not in test8.

This behavior is in both Motif and lesstif.

The Motif version of this test demonstrates what I regard as a bug in
SGI's motif implementation:  Use the Directories Scrolled List to select a
directory without any files, only other directories.  The file list goes
empty and you get "[  ]" in the file list.  After this, there are never
any files in the file list even if you go to a directory that has files!
Why this happens in this test is a mystery, but it happens for test4, but
not test7.  Weird.  LessTif behaves correctly.  Is this a bug in lesstif? :)

==================================================
test9:
Jun 20 1998 JAC

This is a complicated test!
I'm guessing this is one of Rick's babies.
Rick, if you'll tell me what this is supposed to do, I'll test it out.

rws 20 Jun 1998 This is a test of the QualifySearchProc.  You supply it values
in the input structure and it fills in the qualified values in the output
structure.  You can fill in any values on the left hand input structure then
press Qualify, the qualified values should show up on the right hand output
structure.  It also hooks the qualify proc in order to show the input and
output values of the qualify proc.  To do this just play with the FSB in
the middle.

jac 23 Jun 1998
I'm not qualified to evaluate this test, because I haven't dealt that
extensively with FSB semantics.  However, I do note this behavior:

Motif: press filter.  in the input Callbackstruct, value, mask dir and
pattern are all NULL, and the reason and lengths are 0.  Only the event
has a reasonable-looking value.  Press qualify now to get a core dump.

Lesstif: press filter.  The fields of the input callbackstruct are filled
in with reasonable values, and pressing qualify doesn't give a core dump.
This is a bug in lesstif.  We need to core dump :).

==================================================

**********************************************************************
Fixed problems:

test1:
rws 20 Jun 1998 Modified the synthetic resource export functions to grab
the values out of the DirList.  This should fix the dirList part at least.

./test1 > OUT.lesstif
./test1.motif > OUT.motif
diff OUT.lesstif OUT.motif

1c1,4
< c1 c2 c3 0 41 00000029 0 10028f98 10029748
---
> c1 c2 c3 3 41 00000029 10025560 10027780 10027780
> dirList[0] = '/usr/people/jon/src/lesstif/test/Xm/filesb/.'
> dirList[1] = '/usr/people/jon/src/lesstif/test/Xm/filesb/..'
> dirList[2] = '/usr/people/jon/src/lesstif/test/Xm/filesb/CVS'

It appears that the lesstif version is not returning the dirList
resources. rws 20 Jun 1998 should be fixed.

In the Motif version pressing Ok changed the selection field to read "test
it out" This did not happen in the lesstif version.  FIXED: Jun 24 1998,
by rws's patch.

==================================================
test3:
Jun 20 1998 JAC
Motif version wouldn't compile:

/bin/sh ../../../libtool --mode=link cc -g  -g -o test3.motif -o test3.motif test3.o  -L/usr/lib -lXm  -lXt -lX11
cc -g -g -o test3.motif -o test3.motif test3.o -L/usr/lib -lXm -lXt -lX11
ld: ERROR 33: Unresolved text symbol "XdbReason2String" -- 1st referenced by test3.o.
ld: ERROR 33: Unresolved text symbol "XdbXmString2String" -- 1st referenced by test3.o.
ld: INFO 60: Output file removed because of error.
make: *** [test3.motif] Error 1

* rws 20 Jun 1998:
It should compile now. I copied the Xdb routines into the test.

* JAC Jun 23 1998
After the makefile changeover to use the Motif headers, the motif version
won't compile again:

rws 24 Jun 1998 The Motif version should compile now.
jac 24 Jun 1998 Fixed motif compilation errors.

==================================================
test5:

* Fixed Wed Jun 24 1998
lesstif: clicking on the "Button" in the menu bar has no effect
motif: clicking on the "Button" in the menu bar highlights the button.
==================================================

