vista ctp 3508 jit debugger

Hello everyone,
Recently tried Vista CTP Build 3508 32-bit (Beta 2). According to news articles, MS is targeting this CTP for Enterprise customers to start testing with. So here I am :)
The focus of my test is what's new/fixed/etc with the Default JIT (Just In Time) Debugger, which used to be Dr. Watson. The JIT Default Debugger is very important to any developer, as that is the way we can find out about bugs in our code. Yes, there are many great MSFT and 3rd party debuggers out there, and Windows Error Reporting can be used, but many many applications will use the default JIT debugger if they crash.
This help extract from Vista supposedly explains Dr Watson is not being used anymore:
"Dr. Watson, the system failure (or crash) analysis took, has been replaced with Solutions to Problems in Control Panel. This feature helps you check for solutions to common problems that cause Windows or your programs to stop working. If a solution is available, it shows you steps you can take to solve the problem.
You can have Windows automatically report problems and check for solutions or you can choose to check for a solution whenever a problem occurs. Problem descriptions and solutions are saved, so you can view them at any time in Solutions to Problems."
The current big-time beef I have with Dr Watson in XP/2003 is that the Dr Waston log file backtrace, which is very important, is almost useless. The dump can be used, but sometimes users only like to send the log file, or the dump is overwritten by another application crash.
The main reason the backtrace is not usefull is that 99% of the time you cannot trust it, unless OS and application symbols are perfectly setup on the machine. And just who is going to go through all that trouble, for every box/customer that has a crash. That's why symbols servers were invented....
Watson does not support a dynamic symbol server, only a static symbol path, which means you have to download an entire symbol pkg from MSFT, then probably still get the wrong symbols when a hotfix or security fix comes out. This results in the dreaded stack backtrace warning:
WARNING: Stack unwind information not available. Following frames may be wrong.
Which occurs when symbol info is not available for ANY part of the stack. WinDbg and other debuggers seem to be smart enough to insert that statement at the appopriate stack section, but Watson seems to always log it at the top of the stack, even if it has full user symbols for the top of the stack, and doesn't have kernel symbols for the bottom part of the stack.
Ok, so back to Vista, I have an example program that just crashes, despite the registry being set to Dr. Watson (which doesn't even exist in system32)
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\CurrentVersion\AeDebug "Debugger"="drwtsn32 -p %ld -e %ld -g", the JIT debugger is not launched.
Instead I get:
Crash1.exe has stopped working. Windows is checking for a solution to the problem [apparantely this is from the werfault.exe process] [You can only close the program, you don't even a 'click here' for info about the report, such as XP's dwwin.exe] [You can then go to Control Panel -> System and Maintenance -> Problem Reports and Solutions -> View problem history] [Go to Crash1.exe, Right Click, View Problem Details ->] There you see fault address but no backtrace or other info Dr Watson has [You can click 'View a temporary copy of these files] which seems to have memory.hdmp and minidump.mdmp. Full dump and minidump?
You can run werfault.exe at the command line, but can't seem to configure it as you could with Dr Watson.
So a summary of my questions 1. Will Vista JIT Debugger support symbol server technology? Via setting of symbol server path?
2. What enhancements does Solutions to Problems/werfault provide over Dr Watson? Seems better for users expereincing MSFT bugs, but what bugs in all the numerous 3rd party applications, including mine :), loaded on boxes. How can non MSFT developers expect improvements getting that crash data back w/o going through MSFT or implementing HTTP upload to company server (as many customer boxes are behind firewalls and do not even have HTTP access)
3. If a log similar to Dr Watson is created, is there some intelligence in the warning "WARNING: Stack unwind information not available. Following frames may be wrong." if not every single itty bitty symbol file is not present. This is huge problem in not being able to trust watson logs, dumps are OK.
4. Is MSFT going to continue with the policy of shipping outdated, braindead debuggers with the OS. I.e. Why is a much older version of NTSD shipped with XP, not latest version of cdb/windbg/adplus shipped with OS (as comes with Debugging Tools for Windows)
4. Will werfault.exe be configurable? Hopefully much more than Dr Watson but still easy for end-users.
5. Can we expect major changes in the JIT debugger architechture with future Vista beta releases. If not, then I am pretty dissapointed.
Thanks, Ivan Developer with Fortune 500 Company

Zack better answer this one. :-p -- Andre Extended64 | http://www.extended64.com Blog | http://www.extended64.com/blogs/andre http://spaces.msn.com/members/adacosta FAQ for MS AntiSpy http://www.geocities.com/marfer_mvp/FAQ_MSantispy.htm
"Ivan Berg" wrote in message

Hello everyone,
Recently tried Vista CTP Build 3508 32-bit (Beta 2). According to news articles, MS is targeting this CTP for Enterprise customers to start testing with. So here I am :)
The focus of my test is what's new/fixed/etc with the Default JIT (Just In Time) Debugger, which used to be Dr. Watson. The JIT Default Debugger is very important to any developer, as that is the way we can find out about bugs in our code. Yes, there are many great MSFT and 3rd party debuggers out there, and Windows Error Reporting can be used, but many many applications will use the default JIT debugger if they crash.
This help extract from Vista supposedly explains Dr Watson is not being used anymore:
"Dr. Watson, the system failure (or crash) analysis took, has been replaced with Solutions to Problems in Control Panel. This feature helps you check for solutions to common problems that cause Windows or your programs to stop working. If a solution is available, it shows you steps you can take to solve the problem.
You can have Windows automatically report problems and check for solutions or you can choose to check for a solution whenever a problem occurs. Problem descriptions and solutions are saved, so you can view them at any time in Solutions to Problems."
The current big-time beef I have with Dr Watson in XP/2003 is that the Dr Waston log file backtrace, which is very important, is almost useless. The dump can be used, but sometimes users only like to send the log file, or the dump is overwritten by another application crash.
The main reason the backtrace is not usefull is that 99% of the time you cannot trust it, unless OS and application symbols are perfectly setup on the machine. And just who is going to go through all that trouble, for every box/customer that has a crash. That's why symbols servers were invented....
Watson does not support a dynamic symbol server, only a static symbol path, which means you have to download an entire symbol pkg from MSFT, then probably still get the wrong symbols when a hotfix or security fix comes out. This results in the dreaded stack backtrace warning:
WARNING: Stack unwind information not available. Following frames may be wrong.
Which occurs when symbol info is not available for ANY part of the stack. WinDbg and other debuggers seem to be smart enough to insert that statement at the appopriate stack section, but Watson seems to always log it at the top of the stack, even if it has full user symbols for the top of the stack, and doesn't have kernel symbols for the bottom part of the stack.
Ok, so back to Vista, I have an example program that just crashes, despite the registry being set to Dr. Watson (which doesn't even exist in system32)
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\CurrentVersion\AeDebug "Debugger"="drwtsn32 -p %ld -e %ld -g", the JIT debugger is not launched.
Instead I get:
Crash1.exe has stopped working. Windows is checking for a solution to the problem [apparantely this is from the werfault.exe process] [You can only close the program, you don't even a 'click here' for info about the report, such as XP's dwwin.exe] [You can then go to Control Panel -> System and Maintenance -> Problem Reports and Solutions -> View problem history] [Go to Crash1.exe, Right Click, View Problem Details ->] There you see fault address but no backtrace or other info Dr Watson has [You can click 'View a temporary copy of these files] which seems to have memory.hdmp and minidump.mdmp. Full dump and minidump?
You can run werfault.exe at the command line, but can't seem to configure it as you could with Dr Watson.
So a summary of my questions 1. Will Vista JIT Debugger support symbol server technology? Via setting of symbol server path?
2. What enhancements does Solutions to Problems/werfault provide over Dr Watson? Seems better for users expereincing MSFT bugs, but what bugs in all the numerous 3rd party applications, including mine :), loaded on boxes. How can non MSFT developers expect improvements getting that crash data back w/o going through MSFT or implementing HTTP upload to company server (as many customer boxes are behind firewalls and do not even have HTTP access)
3. If a log similar to Dr Watson is created, is there some intelligence in the warning "WARNING: Stack unwind information not available. Following frames may be wrong." if not every single itty bitty symbol file is not present. This is huge problem in not being able to trust watson logs, dumps are OK.
4. Is MSFT going to continue with the policy of shipping outdated, braindead debuggers with the OS. I.e. Why is a much older version of NTSD shipped with XP, not latest version of cdb/windbg/adplus shipped with OS (as comes with Debugging Tools for Windows)
4. Will werfault.exe be configurable? Hopefully much more than Dr Watson but still easy for end-users.
5. Can we expect major changes in the JIT debugger architechture with future Vista beta releases. If not, then I am pretty dissapointed.
Thanks, Ivan Developer with Fortune 500 Company

Woah, I was waiting for you!! :op I'll give it a go *gulp*

So a summary of my questions 1. Will Vista JIT Debugger support symbol server technology? Via setting of symbol server path?

Eventually yes. Windows Server "Longhorn" will have this integrated and working with the latest stuff.

2. What enhancements does Solutions to Problems/werfault provide over Dr Watson? Seems better for users expereincing MSFT bugs, but what bugs in all the numerous 3rd party applications, including mine :), loaded on boxes. How can non MSFT developers expect improvements getting that crash data back w/o going through MSFT or implementing HTTP upload to company server (as many customer boxes are behind firewalls and do not even have HTTP access)

The "Report Problems and Solutions" in the Control Panel is surprisingly good - and it's all done on SSL so it's secure as. I don't think there will be an option to do this, however maybe a server technology like Windows Update Services for Longhorn Server might be able to "fix these" whilst it's there.

3. If a log similar to Dr Watson is created, is there some intelligence in the warning "WARNING: Stack unwind information not available. Following frames may be wrong." if not every single itty bitty symbol file is not present. This is huge problem in not being able to trust watson logs, dumps are OK.

OK there's no question there, however Watson logs and dumps (hehehe) are both secure, and can be trusted. That's like saving a Word document onto your computer and Windows saves it as a virus! That's not going to happen - so don't worry about it.

4. Is MSFT going to continue with the policy of shipping outdated, braindead debuggers with the OS. I.e. Why is a much older version of NTSD shipped with XP, not latest version of cdb/windbg/adplus shipped with OS (as comes with Debugging Tools for Windows)

No word on this yet.

4. Will werfault.exe be configurable? Hopefully much more than Dr Watson but still easy for end-users.

Yep.

5. Can we expect major changes in the JIT debugger architechture with future Vista beta releases. If not, then I am pretty dissapointed.

I presume so, but no official word. And next time, *please* a question at a time!!
-- Zack Whittaker Microsoft Beta (Windows Server R2 Beta Mentor) » ZackNET Enterprises: www.zacknet.co.uk » MSBlog on ResDev: http://msblog.resdev.net » ZackNET Forum: www.zacknet.co.uk/forum » VistaBase: www.zacknet.co.uk/vistabase » This mailing is provided "as is" with no warranties, and confers no rights. All opinions expressed are those of myself unless stated so, and not of my employer, best friend, mother or cat. Let's be clear on that one!
--- Original message follows --- "Andre Da Costa [Extended64]" wrote in message

Zack better answer this one. :-p -- Andre Extended64 | http://www.extended64.com Blog | http://www.extended64.com/blogs/andre http://spaces.msn.com/members/adacosta FAQ for MS AntiSpy http://www.geocities.com/marfer_mvp/FAQ_MSantispy.htm
"Ivan Berg" wrote in message Hello everyone,
Recently tried Vista CTP Build 3508 32-bit (Beta 2). According to news articles, MS is targeting this CTP for Enterprise customers to start testing with. So here I am :)
The focus of my test is what's new/fixed/etc with the Default JIT (Just In Time) Debugger, which used to be Dr. Watson. The JIT Default Debugger is very important to any developer, as that is the way we can find out about bugs in our code. Yes, there are many great MSFT and 3rd party debuggers out there, and Windows Error Reporting can be used, but many many applications will use the default JIT debugger if they crash.
This help extract from Vista supposedly explains Dr Watson is not being used anymore:
"Dr. Watson, the system failure (or crash) analysis took, has been replaced with Solutions to Problems in Control Panel. This feature helps you check for solutions to common problems that cause Windows or your programs to stop working. If a solution is available, it shows you steps you can take to solve the problem.
You can have Windows automatically report problems and check for solutions or you can choose to check for a solution whenever a problem occurs. Problem descriptions and solutions are saved, so you can view them at any time in Solutions to Problems."
The current big-time beef I have with Dr Watson in XP/2003 is that the Dr Waston log file backtrace, which is very important, is almost useless. The dump can be used, but sometimes users only like to send the log file, or the dump is overwritten by another application crash.
The main reason the backtrace is not usefull is that 99% of the time you cannot trust it, unless OS and application symbols are perfectly setup on the machine. And just who is going to go through all that trouble, for every box/customer that has a crash. That's why symbols servers were invented....
Watson does not support a dynamic symbol server, only a static symbol path, which means you have to download an entire symbol pkg from MSFT, then probably still get the wrong symbols when a hotfix or security fix comes out. This results in the dreaded stack backtrace warning:
WARNING: Stack unwind information not available. Following frames may be wrong.
Which occurs when symbol info is not available for ANY part of the stack. WinDbg and other debuggers seem to be smart enough to insert that statement at the appopriate stack section, but Watson seems to always log it at the top of the stack, even if it has full user symbols for the top of the stack, and doesn't have kernel symbols for the bottom part of the stack.
Ok, so back to Vista, I have an example program that just crashes, despite the registry being set to Dr. Watson (which doesn't even exist in system32)
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\CurrentVersion\AeDebug "Debugger"="drwtsn32 -p %ld -e %ld -g", the JIT debugger is not launched.
Instead I get:
Crash1.exe has stopped working. Windows is checking for a solution to the problem [apparantely this is from the werfault.exe process] [You can only close the program, you don't even a 'click here' for info about the report, such as XP's dwwin.exe] [You can then go to Control Panel -> System and Maintenance -> Problem Reports and Solutions -> View problem history] [Go to Crash1.exe, Right Click, View Problem Details ->] There you see fault address but no backtrace or other info Dr Watson has [You can click 'View a temporary copy of these files] which seems to have memory.hdmp and minidump.mdmp. Full dump and minidump?
You can run werfault.exe at the command line, but can't seem to configure it as you could with Dr Watson.
So a summary of my questions 1. Will Vista JIT Debugger support symbol server technology? Via setting of symbol server path?
2. What enhancements does Solutions to Problems/werfault provide over Dr Watson? Seems better for users expereincing MSFT bugs, but what bugs in all the numerous 3rd party applications, including mine :), loaded on boxes. How can non MSFT developers expect improvements getting that crash data back w/o going through MSFT or implementing HTTP upload to company server (as many customer boxes are behind firewalls and do not even have HTTP access)
3. If a log similar to Dr Watson is created, is there some intelligence in the warning "WARNING: Stack unwind information not available. Following frames may be wrong." if not every single itty bitty symbol file is not present. This is huge problem in not being able to trust watson logs, dumps are OK.
4. Is MSFT going to continue with the policy of shipping outdated, braindead debuggers with the OS. I.e. Why is a much older version of NTSD shipped with XP, not latest version of cdb/windbg/adplus shipped with OS (as comes with Debugging Tools for Windows)
4. Will werfault.exe be configurable? Hopefully much more than Dr Watson but still easy for end-users.
5. Can we expect major changes in the JIT debugger architechture with future Vista beta releases. If not, then I am pretty dissapointed.
Thanks, Ivan Developer with Fortune 500 Company

"Zack Whittaker (R2 Mentor)" wrote in message

Woah, I was waiting for you!! :op I'll give it a go *gulp*
So a summary of my questions 1. Will Vista JIT Debugger support symbol server technology? Via setting of symbol server path?
Eventually yes. Windows Server "Longhorn" will have this integrated and working with the latest stuff.

Windows Server Longhorn is currently schedule for release in 2010, correct? Is there some reason why the latest and greatest Vista does not have this built in when other MSFT debuggers have had this for some time?
So this means that for Vista, we are stuck downloading a large symbol pkg from MSFT, which is probably outdated due to hotfixes, to get accurate data of the JIT debugger? This prettty much assures that a download of "Debugging Tools for Windows" will be required then...

2. What enhancements does Solutions to Problems/werfault provide over Dr Watson? Seems better for users expereincing MSFT bugs, but what bugs in all the numerous 3rd party applications, including mine :), loaded on boxes. How can non MSFT developers expect improvements getting that crash data back w/o going through MSFT or implementing HTTP upload to company server (as many customer boxes are behind firewalls and do not even have HTTP access)
The "Report Problems and Solutions" in the Control Panel is surprisingly good - and it's all done on SSL so it's secure as. I don't think there will be an option to do this, however maybe a server technology like Windows Update Services for Longhorn Server might be able to "fix these" whilst it's there.
I suppose I was being overly critical, as we do have other options. It does

seem that "Report Problems and Solutions" is better for end users overall. Perhaps there is a direct way, but I worry that in server-side env, often with no HTTP access to upload crash data, that admins will have to go through a bunch of steps to get at the crash data. Perhaps there is a more direct way though to get at it.

3. If a log similar to Dr Watson is created, is there some intelligence in the warning "WARNING: Stack unwind information not available. Following frames may be wrong." if not every single itty bitty symbol file is not present. This is huge problem in not being able to trust watson logs, dumps are OK.
OK there's no question there, however Watson logs and dumps (hehehe) are both secure, and can be trusted. That's like saving a Word document onto your computer and Windows saves it as a virus! That's not going to happen - so don't worry about it.

Not sure you got my point. Yes the dump files are great and can be looked at offlline. However, the behavior of Dr Watson I mentioned is absolutely true, that warning signifies you cannot trust the stack trace, and often is is wrong. This leads to developers wasting a lot of time looking at code where there is not really a problem. Many ppl were and are confused by this, and I have had many developers show me Dr Watson logs with incorrect stack traces not knowing how incorrect they were.

4. Is MSFT going to continue with the policy of shipping outdated, braindead debuggers with the OS. I.e. Why is a much older version of NTSD shipped with XP, not latest version of cdb/windbg/adplus shipped with OS (as comes with Debugging Tools for Windows)
No word on this yet.
4. Will werfault.exe be configurable? Hopefully much more than Dr Watson but still easy for end-users.
Yep.
Great, any idea how, didn't find anything in the registry or by running

werfault /?. Is this configurability in the current CTP?

5. Can we expect major changes in the JIT debugger architechture with future Vista beta releases. If not, then I am pretty dissapointed.
I presume so, but no official word. And next time, *please* a question at a time!!

Sorry, random brain dumps aren't pretty. LOL

-- Zack Whittaker Microsoft Beta (Windows Server R2 Beta Mentor) » ZackNET Enterprises: www.zacknet.co.uk » MSBlog on ResDev: http://msblog.resdev.net » ZackNET Forum: www.zacknet.co.uk/forum » VistaBase: www.zacknet.co.uk/vistabase » This mailing is provided "as is" with no warranties, and confers no rights. All opinions expressed are those of myself unless stated so, and not of my employer, best friend, mother or cat. Let's be clear on that one!
--- Original message follows --- "Andre Da Costa [Extended64]" wrote in message Zack better answer this one. :-p -- Andre Extended64 | http://www.extended64.com Blog | http://www.extended64.com/blogs/andre http://spaces.msn.com/members/adacosta FAQ for MS AntiSpy http://www.geocities.com/marfer_mvp/FAQ_MSantispy.htm
"Ivan Berg" wrote in message Hello everyone,
Recently tried Vista CTP Build 3508 32-bit (Beta 2). According to news articles, MS is targeting this CTP for Enterprise customers to start testing with. So here I am :)
The focus of my test is what's new/fixed/etc with the Default JIT (Just In Time) Debugger, which used to be Dr. Watson. The JIT Default Debugger is very important to any developer, as that is the way we can find out about bugs in our code. Yes, there are many great MSFT and 3rd party debuggers out there, and Windows Error Reporting can be used, but many many applications will use the default JIT debugger if they crash.
This help extract from Vista supposedly explains Dr Watson is not being used anymore:
"Dr. Watson, the system failure (or crash) analysis took, has been replaced with Solutions to Problems in Control Panel. This feature helps you check for solutions to common problems that cause Windows or your programs to stop working. If a solution is available, it shows you steps you can take to solve the problem.
You can have Windows automatically report problems and check for solutions or you can choose to check for a solution whenever a problem occurs. Problem descriptions and solutions are saved, so you can view them at any time in Solutions to Problems."
The current big-time beef I have with Dr Watson in XP/2003 is that the Dr Waston log file backtrace, which is very important, is almost useless. The dump can be used, but sometimes users only like to send the log file, or the dump is overwritten by another application crash.
The main reason the backtrace is not usefull is that 99% of the time you cannot trust it, unless OS and application symbols are perfectly setup on the machine. And just who is going to go through all that trouble, for every box/customer that has a crash. That's why symbols servers were invented....
Watson does not support a dynamic symbol server, only a static symbol path, which means you have to download an entire symbol pkg from MSFT, then probably still get the wrong symbols when a hotfix or security fix comes out. This results in the dreaded stack backtrace warning:
WARNING: Stack unwind information not available. Following frames may be wrong.
Which occurs when symbol info is not available for ANY part of the stack. WinDbg and other debuggers seem to be smart enough to insert that statement at the appopriate stack section, but Watson seems to always log it at the top of the stack, even if it has full user symbols for the top of the stack, and doesn't have kernel symbols for the bottom part of the stack.
Ok, so back to Vista, I have an example program that just crashes, despite the registry being set to Dr. Watson (which doesn't even exist in system32)
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\CurrentVersion\AeDebug "Debugger"="drwtsn32 -p %ld -e %ld -g", the JIT debugger is not launched.
Instead I get:
Crash1.exe has stopped working. Windows is checking for a solution to the problem [apparantely this is from the werfault.exe process] [You can only close the program, you don't even a 'click here' for info about the report, such as XP's dwwin.exe] [You can then go to Control Panel -> System and Maintenance -> Problem Reports and Solutions -> View problem history] [Go to Crash1.exe, Right Click, View Problem Details ->] There you see fault address but no backtrace or other info Dr Watson has [You can click 'View a temporary copy of these files] which seems to have memory.hdmp and minidump.mdmp. Full dump and minidump?
You can run werfault.exe at the command line, but can't seem to configure it as you could with Dr Watson.
So a summary of my questions 1. Will Vista JIT Debugger support symbol server technology? Via setting of symbol server path?
2. What enhancements does Solutions to Problems/werfault provide over Dr Watson? Seems better for users expereincing MSFT bugs, but what bugs in all the numerous 3rd party applications, including mine :), loaded on boxes. How can non MSFT developers expect improvements getting that crash data back w/o going through MSFT or implementing HTTP upload to company server (as many customer boxes are behind firewalls and do not even have HTTP access)
3. If a log similar to Dr Watson is created, is there some intelligence in the warning "WARNING: Stack unwind information not available. Following frames may be wrong." if not every single itty bitty symbol file is not present. This is huge problem in not being able to trust watson logs, dumps are OK.
4. Is MSFT going to continue with the policy of shipping outdated, braindead debuggers with the OS. I.e. Why is a much older version of NTSD shipped with XP, not latest version of cdb/windbg/adplus shipped with OS (as comes with Debugging Tools for Windows)
4. Will werfault.exe be configurable? Hopefully much more than Dr Watson but still easy for end-users.
5. Can we expect major changes in the JIT debugger architechture with future Vista beta releases. If not, then I am pretty dissapointed.
Thanks, Ivan Developer with Fortune 500 Company

Also, anyone have idea why 2 minidumps are created?
The latest Debugging Tools for Windows windbg (which supports Vista) seems to see them both as minidumps. Dumpchk reports:
<memory.hdmp> 0001 MiniDumpWithDataSegs 0004 MiniDumpWithHandleData 0020 MiniDumpWithUnloadedModules 0100 MiniDumpWithProcessThreadData 0200 MiniDumpWithPrivateReadWriteMemory 0800 MiniDumpWithFullMemoryInfo </memory.hdmp>
<minidump.mdmp> 0001 MiniDumpWithDataSegs 0020 MiniDumpWithUnloadedModules 0100 MiniDumpWithProcessThreadData 0400 MiniDumpWithoutOptionalData </minidump.mdmp>
I would guess so that perhaps the customer could opt-out to provide MSFT with full memory dump for security reasons?
Ivan
"Ivan Berg" wrote in message

Hello everyone,
Recently tried Vista CTP Build 3508 32-bit (Beta 2). According to news articles, MS is targeting this CTP for Enterprise customers to start testing with. So here I am :)
The focus of my test is what's new/fixed/etc with the Default JIT (Just In Time) Debugger, which used to be Dr. Watson. The JIT Default Debugger is very important to any developer, as that is the way we can find out about bugs in our code. Yes, there are many great MSFT and 3rd party debuggers out there, and Windows Error Reporting can be used, but many many applications will use the default JIT debugger if they crash.
This help extract from Vista supposedly explains Dr Watson is not being used anymore:
"Dr. Watson, the system failure (or crash) analysis took, has been replaced with Solutions to Problems in Control Panel. This feature helps you check for solutions to common problems that cause Windows or your programs to stop working. If a solution is available, it shows you steps you can take to solve the problem.
You can have Windows automatically report problems and check for solutions or you can choose to check for a solution whenever a problem occurs. Problem descriptions and solutions are saved, so you can view them at any time in Solutions to Problems."
The current big-time beef I have with Dr Watson in XP/2003 is that the Dr Waston log file backtrace, which is very important, is almost useless. The dump can be used, but sometimes users only like to send the log file, or the dump is overwritten by another application crash.
The main reason the backtrace is not usefull is that 99% of the time you cannot trust it, unless OS and application symbols are perfectly setup on the machine. And just who is going to go through all that trouble, for every box/customer that has a crash. That's why symbols servers were invented....
Watson does not support a dynamic symbol server, only a static symbol path, which means you have to download an entire symbol pkg from MSFT, then probably still get the wrong symbols when a hotfix or security fix comes out. This results in the dreaded stack backtrace warning:
WARNING: Stack unwind information not available. Following frames may be wrong.
Which occurs when symbol info is not available for ANY part of the stack. WinDbg and other debuggers seem to be smart enough to insert that statement at the appopriate stack section, but Watson seems to always log it at the top of the stack, even if it has full user symbols for the top of the stack, and doesn't have kernel symbols for the bottom part of the stack.
Ok, so back to Vista, I have an example program that just crashes, despite the registry being set to Dr. Watson (which doesn't even exist in system32)
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\CurrentVersion\AeDebug "Debugger"="drwtsn32 -p %ld -e %ld -g", the JIT debugger is not launched.
Instead I get:
Crash1.exe has stopped working. Windows is checking for a solution to the problem [apparantely this is from the werfault.exe process] [You can only close the program, you don't even a 'click here' for info about the report, such as XP's dwwin.exe] [You can then go to Control Panel -> System and Maintenance -> Problem Reports and Solutions -> View problem history] [Go to Crash1.exe, Right Click, View Problem Details ->] There you see fault address but no backtrace or other info Dr Watson has [You can click 'View a temporary copy of these files] which seems to have memory.hdmp and minidump.mdmp. Full dump and minidump?
You can run werfault.exe at the command line, but can't seem to configure it as you could with Dr Watson.
So a summary of my questions 1. Will Vista JIT Debugger support symbol server technology? Via setting of symbol server path?
2. What enhancements does Solutions to Problems/werfault provide over Dr Watson? Seems better for users expereincing MSFT bugs, but what bugs in all the numerous 3rd party applications, including mine :), loaded on boxes. How can non MSFT developers expect improvements getting that crash data back w/o going through MSFT or implementing HTTP upload to company server (as many customer boxes are behind firewalls and do not even have HTTP access)
3. If a log similar to Dr Watson is created, is there some intelligence in the warning "WARNING: Stack unwind information not available. Following frames may be wrong." if not every single itty bitty symbol file is not present. This is huge problem in not being able to trust watson logs, dumps are OK.
4. Is MSFT going to continue with the policy of shipping outdated, braindead debuggers with the OS. I.e. Why is a much older version of NTSD shipped with XP, not latest version of cdb/windbg/adplus shipped with OS (as comes with Debugging Tools for Windows)
4. Will werfault.exe be configurable? Hopefully much more than Dr Watson but still easy for end-users.
5. Can we expect major changes in the JIT debugger architechture with future Vista beta releases. If not, then I am pretty dissapointed.
Thanks, Ivan Developer with Fortune 500 Company

Windows Server "Longhorn" is expected just after the release of Windows Vista - around early 2007. That's all I can really answer from there on in... there's only so far my brain will let me wander into the realm of the unsightly and unknown :o)
-- Zack Whittaker Microsoft Beta (Windows Server R2 Beta Mentor) » ZackNET Enterprises: www.zacknet.co.uk » MSBlog on ResDev: http://msblog.resdev.net » ZackNET Forum: www.zacknet.co.uk/forum » VistaBase: www.zacknet.co.uk/vistabase » This mailing is provided "as is" with no warranties, and confers no rights. All opinions expressed are those of myself unless stated so, and not of my employer, best friend, mother or cat. Let's be clear on that one!
--- Original message follows --- "Ivan Berg" wrote in message

"Zack Whittaker (R2 Mentor)" wrote in message Woah, I was waiting for you!! :op I'll give it a go *gulp*
So a summary of my questions 1. Will Vista JIT Debugger support symbol server technology? Via setting of symbol server path?
Eventually yes. Windows Server "Longhorn" will have this integrated and working with the latest stuff.
Windows Server Longhorn is currently schedule for release in 2010, correct? Is there some reason why the latest and greatest Vista does not have this built in when other MSFT debuggers have had this for some time?
So this means that for Vista, we are stuck downloading a large symbol pkg from MSFT, which is probably outdated due to hotfixes, to get accurate data of the JIT debugger? This prettty much assures that a download of "Debugging Tools for Windows" will be required then...
2. What enhancements does Solutions to Problems/werfault provide over Dr Watson? Seems better for users expereincing MSFT bugs, but what bugs in all the numerous 3rd party applications, including mine :), loaded on boxes. How can non MSFT developers expect improvements getting that crash data back w/o going through MSFT or implementing HTTP upload to company server (as many customer boxes are behind firewalls and do not even have HTTP access)
The "Report Problems and Solutions" in the Control Panel is surprisingly good - and it's all done on SSL so it's secure as. I don't think there will be an option to do this, however maybe a server technology like Windows Update Services for Longhorn Server might be able to "fix these" whilst it's there.
I suppose I was being overly critical, as we do have other options. It does seem that "Report Problems and Solutions" is better for end users overall. Perhaps there is a direct way, but I worry that in server-side env, often with no HTTP access to upload crash data, that admins will have to go through a bunch of steps to get at the crash data. Perhaps there is a more direct way though to get at it.
3. If a log similar to Dr Watson is created, is there some intelligence in the warning "WARNING: Stack unwind information not available. Following frames may be wrong." if not every single itty bitty symbol file is not present. This is huge problem in not being able to trust watson logs, dumps are OK.
OK there's no question there, however Watson logs and dumps (hehehe) are both secure, and can be trusted. That's like saving a Word document onto your computer and Windows saves it as a virus! That's not going to happen - so don't worry about it.
Not sure you got my point. Yes the dump files are great and can be looked at offlline. However, the behavior of Dr Watson I mentioned is absolutely true, that warning signifies you cannot trust the stack trace, and often is is wrong. This leads to developers wasting a lot of time looking at code where there is not really a problem. Many ppl were and are confused by this, and I have had many developers show me Dr Watson logs with incorrect stack traces not knowing how incorrect they were.
4. Is MSFT going to continue with the policy of shipping outdated, braindead debuggers with the OS. I.e. Why is a much older version of NTSD shipped with XP, not latest version of cdb/windbg/adplus shipped with OS (as comes with Debugging Tools for Windows)
No word on this yet.
4. Will werfault.exe be configurable? Hopefully much more than Dr Watson but still easy for end-users.
Yep.
Great, any idea how, didn't find anything in the registry or by running werfault /?. Is this configurability in the current CTP?
5. Can we expect major changes in the JIT debugger architechture with future Vista beta releases. If not, then I am pretty dissapointed.
I presume so, but no official word. And next time, *please* a question at a time!!
Sorry, random brain dumps aren't pretty. LOL
-- Zack Whittaker Microsoft Beta (Windows Server R2 Beta Mentor) » ZackNET Enterprises: www.zacknet.co.uk » MSBlog on ResDev: http://msblog.resdev.net » ZackNET Forum: www.zacknet.co.uk/forum » VistaBase: www.zacknet.co.uk/vistabase » This mailing is provided "as is" with no warranties, and confers no rights. All opinions expressed are those of myself unless stated so, and not of my employer, best friend, mother or cat. Let's be clear on that one!
--- Original message follows --- "Andre Da Costa [Extended64]" wrote in message Zack better answer this one. :-p -- Andre Extended64 | http://www.extended64.com Blog | http://www.extended64.com/blogs/andre http://spaces.msn.com/members/adacosta FAQ for MS AntiSpy http://www.geocities.com/marfer_mvp/FAQ_MSantispy.htm
"Ivan Berg" wrote in message Hello everyone,
Recently tried Vista CTP Build 3508 32-bit (Beta 2). According to news articles, MS is targeting this CTP for Enterprise customers to start testing with. So here I am :)
The focus of my test is what's new/fixed/etc with the Default JIT (Just In Time) Debugger, which used to be Dr. Watson. The JIT Default Debugger is very important to any developer, as that is the way we can find out about bugs in our code. Yes, there are many great MSFT and 3rd party debuggers out there, and Windows Error Reporting can be used, but many many applications will use the default JIT debugger if they crash.
This help extract from Vista supposedly explains Dr Watson is not being used anymore:
"Dr. Watson, the system failure (or crash) analysis took, has been replaced with Solutions to Problems in Control Panel. This feature helps you check for solutions to common problems that cause Windows or your programs to stop working. If a solution is available, it shows you steps you can take to solve the problem.
You can have Windows automatically report problems and check for solutions or you can choose to check for a solution whenever a problem occurs. Problem descriptions and solutions are saved, so you can view them at any time in Solutions to Problems."
The current big-time beef I have with Dr Watson in XP/2003 is that the Dr Waston log file backtrace, which is very important, is almost useless. The dump can be used, but sometimes users only like to send the log file, or the dump is overwritten by another application crash.
The main reason the backtrace is not usefull is that 99% of the time you cannot trust it, unless OS and application symbols are perfectly setup on the machine. And just who is going to go through all that trouble, for every box/customer that has a crash. That's why symbols servers were invented....
Watson does not support a dynamic symbol server, only a static symbol path, which means you have to download an entire symbol pkg from MSFT, then probably still get the wrong symbols when a hotfix or security fix comes out. This results in the dreaded stack backtrace warning:
WARNING: Stack unwind information not available. Following frames may be wrong.
Which occurs when symbol info is not available for ANY part of the stack. WinDbg and other debuggers seem to be smart enough to insert that statement at the appopriate stack section, but Watson seems to always log it at the top of the stack, even if it has full user symbols for the top of the stack, and doesn't have kernel symbols for the bottom part of the stack.
Ok, so back to Vista, I have an example program that just crashes, despite the registry being set to Dr. Watson (which doesn't even exist in system32)
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\CurrentVersion\AeDebug "Debugger"="drwtsn32 -p %ld -e %ld -g", the JIT debugger is not launched.
Instead I get:
Crash1.exe has stopped working. Windows is checking for a solution to the problem [apparantely this is from the werfault.exe process] [You can only close the program, you don't even a 'click here' for info about the report, such as XP's dwwin.exe] [You can then go to Control Panel -> System and Maintenance -> Problem Reports and Solutions -> View problem history] [Go to Crash1.exe, Right Click, View Problem Details ->] There you see fault address but no backtrace or other info Dr Watson has [You can click 'View a temporary copy of these files] which seems to have memory.hdmp and minidump.mdmp. Full dump and minidump?
You can run werfault.exe at the command line, but can't seem to configure it as you could with Dr Watson.
So a summary of my questions 1. Will Vista JIT Debugger support symbol server technology? Via setting of symbol server path?
2. What enhancements does Solutions to Problems/werfault provide over Dr Watson? Seems better for users expereincing MSFT bugs, but what bugs in all the numerous 3rd party applications, including mine :), loaded on boxes. How can non MSFT developers expect improvements getting that crash data back w/o going through MSFT or implementing HTTP upload to company server (as many customer boxes are behind firewalls and do not even have HTTP access)
3. If a log similar to Dr Watson is created, is there some intelligence in the warning "WARNING: Stack unwind information not available. Following frames may be wrong." if not every single itty bitty symbol file is not present. This is huge problem in not being able to trust watson logs, dumps are OK.
4. Is MSFT going to continue with the policy of shipping outdated, braindead debuggers with the OS. I.e. Why is a much older version of NTSD shipped with XP, not latest version of cdb/windbg/adplus shipped with OS (as comes with Debugging Tools for Windows)
4. Will werfault.exe be configurable? Hopefully much more than Dr Watson but still easy for end-users.
5. Can we expect major changes in the JIT debugger architechture with future Vista beta releases. If not, then I am pretty dissapointed.
Thanks, Ivan Developer with Fortune 500 Company

Yep, spot on :o)
-- Zack Whittaker Microsoft Beta (Windows Server R2 Beta Mentor) » ZackNET Enterprises: www.zacknet.co.uk » MSBlog on ResDev: http://msblog.resdev.net » ZackNET Forum: www.zacknet.co.uk/forum » VistaBase: www.zacknet.co.uk/vistabase » This mailing is provided "as is" with no warranties, and confers no rights. All opinions expressed are those of myself unless stated so, and not of my employer, best friend, mother or cat. Let's be clear on that one!
--- Original message follows --- "Ivan Berg" wrote in message

Also, anyone have idea why 2 minidumps are created?
The latest Debugging Tools for Windows windbg (which supports Vista) seems to see them both as minidumps. Dumpchk reports:
memory.hdmp 0001 MiniDumpWithDataSegs 0004 MiniDumpWithHandleData 0020 MiniDumpWithUnloadedModules 0100 MiniDumpWithProcessThreadData 0200 MiniDumpWithPrivateReadWriteMemory 0800 MiniDumpWithFullMemoryInfo /memory.hdmp
minidump.mdmp 0001 MiniDumpWithDataSegs 0020 MiniDumpWithUnloadedModules 0100 MiniDumpWithProcessThreadData 0400 MiniDumpWithoutOptionalData /minidump.mdmp
I would guess so that perhaps the customer could opt-out to provide MSFT with full memory dump for security reasons?
Ivan
"Ivan Berg" wrote in message Hello everyone,
Recently tried Vista CTP Build 3508 32-bit (Beta 2). According to news articles, MS is targeting this CTP for Enterprise customers to start testing with. So here I am :)
The focus of my test is what's new/fixed/etc with the Default JIT (Just In Time) Debugger, which used to be Dr. Watson. The JIT Default Debugger is very important to any developer, as that is the way we can find out about bugs in our code. Yes, there are many great MSFT and 3rd party debuggers out there, and Windows Error Reporting can be used, but many many applications will use the default JIT debugger if they crash.
This help extract from Vista supposedly explains Dr Watson is not being used anymore:
"Dr. Watson, the system failure (or crash) analysis took, has been replaced with Solutions to Problems in Control Panel. This feature helps you check for solutions to common problems that cause Windows or your programs to stop working. If a solution is available, it shows you steps you can take to solve the problem.
You can have Windows automatically report problems and check for solutions or you can choose to check for a solution whenever a problem occurs. Problem descriptions and solutions are saved, so you can view them at any time in Solutions to Problems."
The current big-time beef I have with Dr Watson in XP/2003 is that the Dr Waston log file backtrace, which is very important, is almost useless. The dump can be used, but sometimes users only like to send the log file, or the dump is overwritten by another application crash.
The main reason the backtrace is not usefull is that 99% of the time you cannot trust it, unless OS and application symbols are perfectly setup on the machine. And just who is going to go through all that trouble, for every box/customer that has a crash. That's why symbols servers were invented....
Watson does not support a dynamic symbol server, only a static symbol path, which means you have to download an entire symbol pkg from MSFT, then probably still get the wrong symbols when a hotfix or security fix comes out. This results in the dreaded stack backtrace warning:
WARNING: Stack unwind information not available. Following frames may be wrong.
Which occurs when symbol info is not available for ANY part of the stack. WinDbg and other debuggers seem to be smart enough to insert that statement at the appopriate stack section, but Watson seems to always log it at the top of the stack, even if it has full user symbols for the top of the stack, and doesn't have kernel symbols for the bottom part of the stack.
Ok, so back to Vista, I have an example program that just crashes, despite the registry being set to Dr. Watson (which doesn't even exist in system32)
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\CurrentVersion\AeDebug "Debugger"="drwtsn32 -p %ld -e %ld -g", the JIT debugger is not launched.
Instead I get:
Crash1.exe has stopped working. Windows is checking for a solution to the problem [apparantely this is from the werfault.exe process] [You can only close the program, you don't even a 'click here' for info about the report, such as XP's dwwin.exe] [You can then go to Control Panel -> System and Maintenance -> Problem Reports and Solutions -> View problem history] [Go to Crash1.exe, Right Click, View Problem Details ->] There you see fault address but no backtrace or other info Dr Watson has [You can click 'View a temporary copy of these files] which seems to have memory.hdmp and minidump.mdmp. Full dump and minidump?
You can run werfault.exe at the command line, but can't seem to configure it as you could with Dr Watson.
So a summary of my questions 1. Will Vista JIT Debugger support symbol server technology? Via setting of symbol server path?
2. What enhancements does Solutions to Problems/werfault provide over Dr Watson? Seems better for users expereincing MSFT bugs, but what bugs in all the numerous 3rd party applications, including mine :), loaded on boxes. How can non MSFT developers expect improvements getting that crash data back w/o going through MSFT or implementing HTTP upload to company server (as many customer boxes are behind firewalls and do not even have HTTP access)
3. If a log similar to Dr Watson is created, is there some intelligence in the warning "WARNING: Stack unwind information not available. Following frames may be wrong." if not every single itty bitty symbol file is not present. This is huge problem in not being able to trust watson logs, dumps are OK.
4. Is MSFT going to continue with the policy of shipping outdated, braindead debuggers with the OS. I.e. Why is a much older version of NTSD shipped with XP, not latest version of cdb/windbg/adplus shipped with OS (as comes with Debugging Tools for Windows)
4. Will werfault.exe be configurable? Hopefully much more than Dr Watson but still easy for end-users.
5. Can we expect major changes in the JIT debugger architechture with future Vista beta releases. If not, then I am pretty dissapointed.
Thanks, Ivan Developer with Fortune 500 Company

This is also in XP today. -- -- Andre Windows Connect | http://www.windowsconnected.com Extended64 | http://www.extended64.com Blog | http://www.extended64.com/blogs/andre http://spaces.msn.com/members/adacosta FAQ for MS AntiSpy http://www.geocities.com/marfer_mvp/FAQ_MSantispy.htm "Ivan Berg" wrote in message

Also, anyone have idea why 2 minidumps are created?
The latest Debugging Tools for Windows windbg (which supports Vista) seems to see them both as minidumps. Dumpchk reports:
memory.hdmp 0001 MiniDumpWithDataSegs 0004 MiniDumpWithHandleData 0020 MiniDumpWithUnloadedModules 0100 MiniDumpWithProcessThreadData 0200 MiniDumpWithPrivateReadWriteMemory 0800 MiniDumpWithFullMemoryInfo /memory.hdmp
minidump.mdmp 0001 MiniDumpWithDataSegs 0020 MiniDumpWithUnloadedModules 0100 MiniDumpWithProcessThreadData 0400 MiniDumpWithoutOptionalData /minidump.mdmp
I would guess so that perhaps the customer could opt-out to provide MSFT with full memory dump for security reasons?
Ivan
"Ivan Berg" wrote in message Hello everyone,
Recently tried Vista CTP Build 3508 32-bit (Beta 2). According to news articles, MS is targeting this CTP for Enterprise customers to start testing with. So here I am :)
The focus of my test is what's new/fixed/etc with the Default JIT (Just In Time) Debugger, which used to be Dr. Watson. The JIT Default Debugger is very important to any developer, as that is the way we can find out about bugs in our code. Yes, there are many great MSFT and 3rd party debuggers out there, and Windows Error Reporting can be used, but many many applications will use the default JIT debugger if they crash.
This help extract from Vista supposedly explains Dr Watson is not being used anymore:
"Dr. Watson, the system failure (or crash) analysis took, has been replaced with Solutions to Problems in Control Panel. This feature helps you check for solutions to common problems that cause Windows or your programs to stop working. If a solution is available, it shows you steps you can take to solve the problem.
You can have Windows automatically report problems and check for solutions or you can choose to check for a solution whenever a problem occurs. Problem descriptions and solutions are saved, so you can view them at any time in Solutions to Problems."
The current big-time beef I have with Dr Watson in XP/2003 is that the Dr Waston log file backtrace, which is very important, is almost useless. The dump can be used, but sometimes users only like to send the log file, or the dump is overwritten by another application crash.
The main reason the backtrace is not usefull is that 99% of the time you cannot trust it, unless OS and application symbols are perfectly setup on the machine. And just who is going to go through all that trouble, for every box/customer that has a crash. That's why symbols servers were invented....
Watson does not support a dynamic symbol server, only a static symbol path, which means you have to download an entire symbol pkg from MSFT, then probably still get the wrong symbols when a hotfix or security fix comes out. This results in the dreaded stack backtrace warning:
WARNING: Stack unwind information not available. Following frames may be wrong.
Which occurs when symbol info is not available for ANY part of the stack. WinDbg and other debuggers seem to be smart enough to insert that statement at the appopriate stack section, but Watson seems to always log it at the top of the stack, even if it has full user symbols for the top of the stack, and doesn't have kernel symbols for the bottom part of the stack.
Ok, so back to Vista, I have an example program that just crashes, despite the registry being set to Dr. Watson (which doesn't even exist in system32)
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\CurrentVersion\AeDebug "Debugger"="drwtsn32 -p %ld -e %ld -g", the JIT debugger is not launched.
Instead I get:
Crash1.exe has stopped working. Windows is checking for a solution to the problem [apparantely this is from the werfault.exe process] [You can only close the program, you don't even a 'click here' for info about the report, such as XP's dwwin.exe] [You can then go to Control Panel -> System and Maintenance -> Problem Reports and Solutions -> View problem history] [Go to Crash1.exe, Right Click, View Problem Details ->] There you see fault address but no backtrace or other info Dr Watson has [You can click 'View a temporary copy of these files] which seems to have memory.hdmp and minidump.mdmp. Full dump and minidump?
You can run werfault.exe at the command line, but can't seem to configure it as you could with Dr Watson.
So a summary of my questions 1. Will Vista JIT Debugger support symbol server technology? Via setting of symbol server path?
2. What enhancements does Solutions to Problems/werfault provide over Dr Watson? Seems better for users expereincing MSFT bugs, but what bugs in all the numerous 3rd party applications, including mine :), loaded on boxes. How can non MSFT developers expect improvements getting that crash data back w/o going through MSFT or implementing HTTP upload to company server (as many customer boxes are behind firewalls and do not even have HTTP access)
3. If a log similar to Dr Watson is created, is there some intelligence in the warning "WARNING: Stack unwind information not available. Following frames may be wrong." if not every single itty bitty symbol file is not present. This is huge problem in not being able to trust watson logs, dumps are OK.
4. Is MSFT going to continue with the policy of shipping outdated, braindead debuggers with the OS. I.e. Why is a much older version of NTSD shipped with XP, not latest version of cdb/windbg/adplus shipped with OS (as comes with Debugging Tools for Windows)
4. Will werfault.exe be configurable? Hopefully much more than Dr Watson but still easy for end-users.
5. Can we expect major changes in the JIT debugger architechture with future Vista beta releases. If not, then I am pretty dissapointed.
Thanks, Ivan Developer with Fortune 500 Company