[M3devel] Juno on NT (presumably canary for other problems)

[Jay K]

> unless ya'll already have that covered on a VM for Hudson

 
We do. Another wouldn't hurt much though, except time/energy.
 
 
Maybe document the setup.
 
 
We had a bunch of problems, though some were due to wierdo missteps that I would have called out had I been paying close attention. That is, setup is more like "leave the defaults, don't do extra wierd stuff".
 
 
CVSNT by default changes line endings, breaking all .sh files.
  Maybe there is a setting to fix it? I'm not sure. The documentation is very confusing on this point. It talks about a keyword setting on a per file basis. Or maybe a checkout/update option?
 
 
So Cygwin text mode mounts were used, which partly fixes the problem, but leaves caltech-parser broken (it should be fixed to allow for \n, \r, or \r\n.).
Had I noticed I would have avoided the text mode mount in the first place.
 
 
Fixed by:
 Use Cygwin cvs as I've been using all along.
 Don't use textmode mounts, use the default.
 
 
Also cygwin mount cygexec was being used. I don't know what that really does.
No longer used. It's not the default.
 
 
Cygwin sshd has a bug that breaks Visual C++ link.exe. Seriously.
We switched to another sshd, and paid a small licensing fee.
 
 
Sometimes reboot needed to get sshd/ssh/scp working.
 
 
The initial setup up cm3 itself has some kinks as well that Olaf pledges to improve after the current release -- where to find a baseline cm3 to start with and possibly m3core/libm3.
 
 
It's probably easier, though still a small pain, to run the Tinderbox stuff.
I checked in a document detailing my experience.
Tinderbox doesn't require Java or sshd.
I would still avoid CVSNT unless that behavior can be remedied.
 
 
 
 - Jay



________________________________
> Date: Fri, 4 Sep 2009 16:21:47 -0400
> From: rcoleburn at scires.com
> To: m3devel at elegosoft.com
> Subject: Re: [M3devel] Juno on NT (presumably canary for other problems)
>
>
>
>
>
>
>
>
>
> Jay:
>
>
>
> Yeah I see the Juno window start up on most runs, but then it crashes so quickly that you can't see much before the window disappears.
>
>
>
> I use formsedit quite a bit. I recall that back when we purchased Reactor/cm3 v4.1, that Farshad and company had to make some patches to formsedit because it crashed on certain platforms. Critical Mass provided us with the patched executables, but I never got the sources. Not sure if the patches got put into what we have now.
>
>
>
> There are some problems now with formsedit that I've documented in the past. One of the big ones has to deal with keeping the cursor placement in sync between what is shown on the display and what the program actually thinks, particularly when switching between mouse and keyboard movement.
>
>
>
> Similarly, I know that Reactor was a work in progress. Although I worked with Farshad to obtain the sources for the open source release as CM3IDE, I'm not confident that all the latest greatest sources actually made it to me in the version Farshad supplied. He had to dig back through some archives to obtain them. I've noticed some operational differences between my old "Reactor" and the new "CM3IDE".
>
>
>
> Monday is a holiday for me, so I'll try to devote some time this long weekend on running more tests. If you want me to run stuff on HP-UX, I'll have to pull the machine "out-of-moth-balls" so to speak because its been a long time since I've used it. I think I still have three HP-UX machines. One of them had a hard drive problem with one of its SCSI drives so that may take some to repair. I'll also try to look at some of the scripts ya'll are using for the automated tests and see if there is a way I can run some of these tests on XP and Vista (unless ya'll already have that covered on a VM for Hudson--let me know if I don't need to do this).
>
>
>
> Regards,
>
> Randy
>
>>>> Jay K 9/4/2009 3:44 PM>>>
> Randy, thanks, understood. Actually Juno does draw a bunch of fancy stuff before the assertion failure. And formsedit at least historically works, and I thought mentor but I'll have to retest after Tony fixes, and with old versions. Yes please test anything you can. I can get you an HP-UX toolset if you need..assumiing I can resolve why last I powered the machine on I couldn't see it on the network. At some point we might even get HP-UX in Hudson. :)
>
> - Jay
>
>
>
>
> ________________________________
>
>
> Date: Fri, 4 Sep 2009 11:10:53 -0400
> From: rcoleburn at scires.com
> To: jay.krell at cornell.edu; m3devel at elegosoft.com
> Subject: Re: [M3devel] Juno on NT (presumably canary for other problems)
>
>
>
>
> Jay said:
>
>>This we shall blame on Trestle not fully being ported to Win32, I guess.
>>At the very least, it seems to the behavior going back a while.
>>You can occasionally see this in head, but usually you see #3.
>
> I think there are some things that aren't fully ported as you say, but I have been using Trestle and FormsVBT on Windows since the mid nineties. My company paid Critical Mass to get Trestle/FormsVBT working on Windows so we could write code whose GUI behaved the same on both HPUX and WindowsNT. I am pretty sure that all those changes got factored into what we have now. At least, I am able to build my GUI apps and they do run on XP and Vista. Granted, I haven't done extensive tests with them using latest sources. Maybe I should do some checking here.
>
>
>
> Regards,
>
> Randy