chaat
Sophomore Member
Posts: 59
|
Post by chaat on Feb 1, 2015 12:53:13 GMT -5
Hi, Last night I was doing some coding on a COBOL program. I had 3 instances of SPFLite open at the time. One for editing, one to look at the output compiler listing and a third to browse another program from which i was copying blocks of code.
During this time, the session which was doing the editing crashed. My question is when that happens is any dump created which might help Roger with problem determination. I'm very familiar with using dumps to do problem determination on the mainframe environment (Z/OS). Does something similar exist on the windows environment for application programs.
When SPFLite crashes that results in a loss of work when in the midst of an editing session. Last year I asked about automating VSAVE to provide functionality similar to what RECOVERY ON does in ISPF.
Anyway I'm starting to ramble.
If there is a dump file, could you post information about how to locate it and send it to you ?
if not, is that something that you could add so that it would produce a dump file when the program crashes.
Thanks
Chuck H.
|
|
chaat
Sophomore Member
Posts: 59
|
Post by chaat on Feb 2, 2015 16:08:22 GMT -5
i see a file from earlier, 09/16/2014 but not for this last week when the crash occurred. perhaps there are some crashes which you don't get a chance to capture.
|
|
|
Post by George on Feb 3, 2015 12:32:42 GMT -5
chaat: Robert:
Debugging crashes is tough. SPFLite is written in PowerBasic, and you basically have the source and you have the EXE. For old mainframe types like us, there is nothing whatsoever from the compiler about the EXE. No generated ASM code, no memory maps, no Link-Edit maps, nada.
So when windows sometime pops up with all kinds of lovely hex addresses and mini portions of memory displayed, it is totally and completely useless. Picture a mainframe dump with no compiler listing, no link-edit map, just the raw source code.
That's why I'm always asking for details on the sequence of events leading to the crash, a repeatable series of steps to cause it, etc. The Callback trace that SPFLite tries to produce is the best I can get, at least it tells me roughly what area of code failed.
So Robert's reference to C:\Users\<username>\Documents\SPFLite\SPFLiteCrash.txt is correct. It's the only thing useful. But a repeatable scenario to create the crash is simply Gold, I can almost guarantee those can be found and fixed.
George
|
|
|
Post by George on Feb 3, 2015 15:06:43 GMT -5
chaat: Regarding the automatic VSAVE stuff. I've had a request from another user for automatic SAVE. Would that do it? Or only VSAVE so you can retain the original file at it's starting point. I started some changes for this, maybe I could make it handle either type of save.
Comments?
George
|
|
chaat
Sophomore Member
Posts: 59
|
Post by chaat on Feb 5, 2015 18:25:06 GMT -5
for me, i would prefer the vsave as there are times when i've messed up a big block of code and want to abandon all changes and start over. vsave would allow that while still offering functionality somewhat like recovery in ispf.
|
|
|
Post by George on Feb 6, 2015 13:06:33 GMT -5
chaat: OK, I've made it an option, so you can choose SAVE or VSAVE. When I have a test version to play with, I'll repost the news here.
George
|
|
chaat
Sophomore Member
Posts: 59
|
Post by chaat on Feb 7, 2015 14:06:56 GMT -5
thanks George
|
|
|
Post by George on Feb 9, 2015 12:02:22 GMT -5
chaat: If you interested, here's a test version www.spflite.com/Files/SPFLite815040.ex_ Just rename to SPFLite.EXE and swap into the instal folder. (Maybe rename the Prod. one first, just in case) You can request the new option either by a PROF EDIT profname command (enter n,[SAVE|VSAVE] in the input box), or while editing a file using the Profile with the command ACTIONSAVE n,[SAVE|VSAVE] (Make sure the profile is unlocked to save the value.) the 'n' operand is the frequency in terms of 'interactions which alter the data'. e.g. an interaction that does some line commands, and primary change type commands at the same time is ONE interaction. Think of it as # times you press ENTER or a PFK command and data is modified. Comments appreciated. George
|
|
|
Post by George on Feb 10, 2015 10:57:16 GMT -5
Robert: Yes, originally there wasn't a SAVE/VSAVE option and it made a bit more sense. I'll alter that before a general release.
George
|
|