T.R | Title | User | Personal Name | Date | Lines |
---|
437.1 | | HYDRA::CHIN | | Wed Dec 18 1996 09:17 | 22 |
437.2 | | DECCXL::OUELLETTE | | Wed Dec 18 1996 09:58 | 7 |
437.3 | | HYDRA::CHIN | | Wed Dec 18 1996 16:10 | 8 |
437.4 | | HYDRA::CHIN | | Thu Jan 02 1997 14:17 | 17 |
437.5 | existing workaround is safe | DECWET::PETERSON | | Fri Jan 03 1997 14:33 | 21 |
437.6 | How do they do the "link dump" ? | DECWET::MUPOPA | Glorious absence of sophistication | Mon Jan 06 1997 11:52 | 13 |
437.7 | | HYDRA::CHIN | | Mon Jan 06 1997 14:48 | 25 |
437.8 | | DECWET::LUNAS | | Mon Jan 13 1997 11:19 | 15 |
437.9 | | HYDRA::AMORELLI | | Thu Feb 06 1997 17:24 | 43 |
| I'm now helping Miller on this account.
A couple of things. First, the work-around (linkdump) is completely
unacceptable to them. And, their attitude is it works on Intel better
and FASTER, it should work as well and FASTER on Alpha. It doesn't! it's NOT!
I have seen the problem (and other problems as well - later) and it happens
almost everytime. It never fails on the Intel system. The error can be
reproduced quite easily. Unfortunately the condition seems to be context
dependent. At this point they are unwilling to release the source code
for testing purposes. Perhaps I can create a reproduce on-site next week.
They don't alter the environment settings during a given session.
Here's the c:\linkdump\linkdump and switches when the session fails. I've
asked for the makefile. If you think it'll help let me know, it's kinda big.
==========================
link.rsp
/nologo
/subsystem:windows
/dll
/incremental:yes
/pdb:".\sldappud.pdb"
/debug
/machine:ALPHA
/nodefaultlib
/out:".\sldappud.dll"
/implib:WinDebug/sldappud.lib
==========================
Here is the stuff from the project settings link tab:
=================================================
mfc42ud.lib mfco42ud.lib mfcs42ud.lib msvcrtd.lib KERNEL32.LIB
advapi32.lib uuid.lib uuid2.lib user32.lib shell32.lib GDI32.LIB
oleaut32.lib ole32.lib comctl32.lib /nologo /subsystem:windows /dll
/incremental:yes /pdb:"../../../Alpha/WinDebug/sldappud.pdb" /debug
/machine:ALPHA /nodefaultlib /out:"../../../Alpha/WinDebug/sldappud.dll"
/implib:"WinDebug/sldappud.lib"
======================================
|
437.10 | | DECWET::THOMAS | Bug-for-bug compatible with Intel | Tue Feb 11 1997 16:36 | 16 |
| Creating the linkdump by setting LINK_REPRO wasn't intended to be a
workaround, it was intended to get more information for us, but it
needs to be provided from a link that fails (which apparently isn't the
case here).
Since we can't get a repro case from Solidworks, can you persuade them
to try this under VC 5.0 beta 2? This will at least give us the
opportunity to fix this in 5.0.
If it is a problem in 5.0 beta 2, please report it in
DECWET::VISUAL_FT.
I understand Solidworks is a beta site; we got our CD today, so they
should have theirs soon.
Thanks ... Mike
|
437.11 | | HYDRA::AMORELLI | | Thu Feb 13 1997 16:53 | 14 |
| Hi Mike,
I called solidworks about the linkdump file in .9 and was told that was from
a failed link. Make sense? Does this mean anything?
You ask about VC++ 5.0 beta 2. Umm? Tomorrow I'll ask if they're willing to
try it. I know they have it. In fact, I'll do the install and try it myself.
I'll let you know how it goes.
Do you have any other suggestions at this point?
Thanks,
Carl
|
437.12 | | NETRIX::"[email protected]" | Carl Amorelli | Mon Feb 17 1997 10:25 | 137 |
| Here are the link dumps, one bad, one good. I didn't have time
to try VC++ 5.0.
------------------------
Failure:
/nologo
/subsystem:windows
/dll
/incremental:yes
/pdb:".\sldappud.pdb"
/debug
/machine:ALPHA
/nodefaultlib
/out:".\sldappud.dll"
/implib:WinDebug/sldappud.lib
----------------------------------------
Success:
/nologo
/subsystem:windows
/dll
/incremental:yes
/pdb:".\sldappud.pdb"
/debug
/machine:ALPHA
/nodefaultlib
/out:".\sldappud.dll"
/implib:WinDebug/sldappud.lib
".\mfc42ud.lib"
".\mfco42ud.lib"
".\mfcs42ud.lib"
".\msvcrtd.lib"
".\KERNEL32.LIB"
".\advapi32.lib"
".\uuid.lib"
".\uuid2.lib"
".\user32.lib"
".\shell32.lib"
".\GDI32.LIB"
".\oleaut32.lib"
".\ole32.lib"
".\comctl32.lib"
".\dsymtable.obj"
".\sldhasp.obj"
".\constrtb.obj"
".\selfilter.obj"
".\macro.obj"
".\customize.obj"
".\mainfrm.obj"
".\seltb.obj"
".\userjobdlg.obj"
".\ampage.obj"
".\selcbox.obj"
".\uiregistry.obj"
".\uicancelmessage.obj"
".\basecmd.obj"
".\DLGUTILS.OBJ"
".\viewtb.obj"
".\FONTDLG.OBJ"
".\errordlg.obj"
".\uidialog.obj"
".\uitemplate.obj"
".\arutils.obj"
".\undo.obj"
".\mtextdlg.obj"
".\assemtb.obj"
".\userjobmgr.obj"
".\haspntms.obj"
".\toolbarctrl.obj"
".\sldapp.obj"
".\baseview.obj"
".\skmaintb.obj"
".\uipageheaderfooterdlg_c.obj"
".\auframe.obj"
".\environment.obj"
".\AUEnvironment_C.obj"
".\filparse.obj"
".\dsymbols.obj"
".\maintb.obj"
".\drawtb.obj"
".\linefontdef.obj"
".\SUMMINFO.OBJ"
".\cmdman.obj"
".\basedocmaps.obj"
".\baserectitem.obj"
".\basedoc.obj"
".\am.obj"
".\macrotb.obj"
".\toolbar.obj"
".\customizemenu.obj"
".\uoprint.obj"
".\dllhusk.obj"
".\customizekey.obj"
".\dlgcmd.obj"
".\SUMMPAGE.OBJ"
".\dsym.obj"
".\printdlg.obj"
".\uicustomheaderfooterdlg_c.obj"
".\uipage.obj"
".\PROPSET.OBJ"
".\memmonitor.obj"
".\appfontarch.obj"
".\refenviron.obj"
".\uimdoctempl.obj"
".\undo_command.obj"
".\apputils.obj"
".\uidocmgr.obj"
".\cmdline.obj"
".\uiutils.obj"
".\auam_c.obj"
".\BLOCKMAN.OBJ"
".\ddata.obj"
".\recorder.obj"
".\constraint.obj"
".\featnettb.obj"
".\apppch.obj"
".\tbdlg.obj"
".\UI3DCONN.OBJ"
".\briconmap.obj"
".\sketchtb.obj"
".\envxhatch.obj"
".\tbcmd.obj"
".\dropeffect.obj"
".\registerdoc.obj"
".\linestyle.obj"
".\ipframe.obj"
".\basemode.obj"
".\sldmgud.lib"
".\sldgcud.lib"
".\sldebud.lib"
".\sldutud.lib"
".\TESTDLL2.lib"
".\sldmfcud.lib"
[Posted by WWW Notes gateway]
|
437.13 | comments from a bystander | DECC::OUELLETTE | | Tue Feb 18 1997 15:29 | 2 |
| You'll need to suply all of the files mentioned in the link dump.
You'll need to collect those after a bad dump from the special directory.
|
437.14 | Wait for a release candidate kit for VC++ 5.0 | DECWET::MUPOPA | | Tue Feb 18 1997 17:23 | 5 |
| We have a release candidate VC++ 5.0 kit planned for the end
of this week that we would like you to try.
Several bugs have been fixed in the debugger (single step,
watchable local variables, etc) and the incremental linker.
|
437.15 | No way to .13 & way to .14 | HYDRA::AMORELLI | | Thu Feb 20 1997 14:10 | 13 |
| Re: .13
There's no way Solidworks will release any of the files from the linkdump. I've
tried to convince them in order to reproduce the error we must have the files.
The simply will not release any part of their product to anyone. I have given up
on this issue. Sorry.
Re:.14
I'll take the VC++ 5.0 release candidate kit when it's available. Just let me
know when and where to pick it up.
Thanks.
|