Sb7 ide gpf

[NewsArchive]

Friedrich,

I have a script that compiles, but when signing the install, it gets an error code
2 and the IDE always crashes at this point. Can replicate every time. So I
changed the time stamp URL from http://timestamp.comodoca.com/authenticode to
http://timestamp.verisign.com/scripts/timstamp.dll and I got it through with no
problems.

I'm theorizing that the Comodo time stamp was not available at the time I
compiled. So if it signs for you, then perhaps you need to go offline to prevent
the time stamp from working. Then you should be able to replicate the problem.

HTH,
--
Russell B. Eggen
www.radfusion.com
Clarion developers: www.radfusion.com/devs.htm

[NewsArchive]

Version: 7.2.3015

--
Russell B. Eggen
www.radfusion.com
Clarion developers: www.radfusion.com/devs.htm

[NewsArchive]

Hi Russ,

>
> Version: 7.2.3015
>

Comodo did a server update a few hours ago (it's back online now).
Sometimes, Signcode.exe crashes the IDE if the timestamp server is not
available or if the code-sign password is incorrect.

Friedrich

[NewsArchive]

>>
>> Version: 7.2.3015
>>

BTW, this is not version specific. It can happen with all SB versions.

Friedrich

[NewsArchive]

OK - you know how this works, if the version is not supplied, then you need it.
Otherwise .. <g>

But the code sign failure is not the issue (it happens), it's the IDE crashing
after the error is reported.

--
Russell B. Eggen
www.radfusion.com
Clarion developers: www.radfusion.com/devs.htm

[NewsArchive]

Russ not sure if this helps.

But I get this quite a lot where the timestamp server is down. But I have
never had the IDE crash on any version of SB including the latest. It just
stops with an error as expected. OK it does not report exaclty what the
problem is but I know by now what it is and wait a few minutes and try
again.

John Fligg

[NewsArchive]

John,

I know sometimes the server is not available and the error reported by SB is fine
(and SB is behaving correctly here). I've learned that the error code means that
the server is not available. I'm not shooting the messenger here <g>.

There are times when a compiler error happens, the SB 7 IDE crashes. I was
compiling this script, and the crash was 100% reproducible (tried 5 times <g>).
Of course, there are times when I get an error and the IDE is fine. This was not
one of those times.

I worked around the crash by choosing the other server. That told me the original
server was down.

--
Russell B. Eggen
www.radfusion.com
Clarion developers: www.radfusion.com/devs.htm

[NewsArchive]

> But the code sign failure is not the issue (it happens), it's the IDE
> crashing after the error is reported.

Yes. The IDE waits for the signtool executable and if signtool fails
(timestamp server down or wrong code-sign password) then the this corrupts
the IDE stack and it GPFs. Happens with all SB versions.

Friedrich

[NewsArchive]

Hi Friedrich,

this could be the same problem that I have and Jane confirmed:

news://news.softvelocity.com:119/A921.1279122493.9346@discuss.softvelocity.com

This has nothing to do with signcode.exe, because we have this problem
when we use gibtesnicht.exe

Markus

[NewsArchive]

Hi Markus,

> this could be the same problem that I have and Jane confirmed:
>
> news://news.softvelocity.com:119/A921.1279122493.9346@discuss.softvelocity.com
>
> This has nothing to do with signcode.exe, because we have this problem
> when we use gibtesnicht.exe

Unfortunately, the issues are not related :-(

Friedrich

[NewsArchive]

We have debugged the IDE and I think it is caused by the PKEY.EXE
application which tries to "inject" the code-signature password into the
SignCode (SPC/PVK) password window. This injection process seems to corrupt
the sbuilder.exe stack.

The same does not happen when using SignTool.exe -- PKEY.EXE is not required
here because SignTool (PFX) supports passwords via the command line.

I don't know if it is possible to fix this.

Friedrich

[NewsArchive]

Should read:

"...This injection process seems to corrupt the sbuilder.exe (in SB6) and
sb7.exe (in SB7) stack..."

Friedrich

[NewsArchive]

Sounds right. I've no knowledge how you can dump a corrupt stack, I'll leave that
puzzle to the real low level bit twiddlers (if its even possible) <g>.

But if the timecode logic comes back with an error, perhaps abort the compilation
right there so PKEY.EXE is never called?

--
Russell B. Eggen
www.radfusion.com
Clarion developers: www.radfusion.com/devs.htm

[NewsArchive]

Hi Russ,

> Sounds right. I've no knowledge how you can dump a corrupt stack, I'll
> leave that puzzle to the real low level bit twiddlers (if its even
> possible) <g>.
>
> But if the timecode logic comes back with an error, perhaps abort the
> compilation right there so PKEY.EXE is never called?

The problem is that PKEY.EXE has to be launched before SignCode.exe is
started. Otherwise, it would not be possible to inject the password into
the SignCode dialog.

I fear, the only real solution is to switch to a PFX certificate format.

Friedrich

[NewsArchive]

OK. I've got a pretty good grasp of the issue here. Also a decent workaround so
I'm not stopped if it ever happens again. Appreciate the feedback.

--
Russell B. Eggen
www.radfusion.com
Clarion developers: www.radfusion.com/devs.htm

[NewsArchive]

OK. That makes sense. Its not a big deal since the script and project are saved
before compiling.

--
Russell B. Eggen
www.radfusion.com
Clarion developers: www.radfusion.com/devs.htm

[NewsArchive]

>
> Unfortunately, the issues are not related :-(
>

The above is incorrect -- the issues *are* related. If a SPC/PVK
code-signing (with password) process fails (for whatever reason), the
external tool PKEY.EXE can crash the IDE.

Friedrich