|
Post by Jo on Jul 15, 2018 5:56:43 GMT -5
Sometimes I experienced SPFlite Crash after PF3 (End) on one of the Edit-Tabs. Since the member was sucsessfully saved, and I usually don't remember what was before, I did not report this crash earlier. But now I searched for the Crash-Report, but there was none. No SPFLiteCrash....-File for this incident. So I looked for a Screen-Copy-Program (I was sure, I had one), but in the meadtime SPFLite and the Crashreport-Window disappeared... So, what do I have? I remember: The exception was C0000005, KB Primitive: ENTER, Line Cmnd: C I see: The member was saved, the ...BKP...-File was created (Autobkup was ON), the "Recent Files.FLIST" and "Recent Paths.FLIST" are saved with same Timestamp as the member, the ...STATE-File was saved with same Timestamp as the member.
I hope, I'm not the only one seeing that sort of crash someone has a better doc.
Sorry, Jo
|
|
|
Post by George on Jul 15, 2018 9:31:41 GMT -5
Jo, The C0000005 is just one of those "memory access violation" errors, like saying "You had a problem". Even the crash reports themselves are of little use.
Hi level languages are great, but I miss the days of doing everything in Assembler and crashes/dumps/etc. were actual pointers back into the original code. Now, Windows likes to dump all kinds of pointers, memory addresses etc. out and all you can really do is shrug and stare at original hi-level code. It's very frustrating for us all, but a repetitive failure sequence to trigger the error is whats needed.
George
|
|
|
Post by Jo on Jul 15, 2018 16:39:27 GMT -5
George, yes, I know that these type of non-reproducible errors are most annoying and always are hard to detect. I have one more problem that I cannot reproduce: When I select a member from say the Found-Filelist, that's already in an open tab, I normally get that tab in the foreground and the message "File already open in this tab". But sometimes there is only a beep, an I stay in FM. And sometimes another edit-tab is opened with the same member. I could not manage to reproduce any of these errors. So, what can I do to collect a useful error-doc? Shall I install a keylogger? Or can you reactivate that extensive trace.logging from 8.2 three years ago?
Jo
|
|
|
Post by George on Jul 16, 2018 10:28:19 GMT -5
Jo: That old trace logging I am slowly removing as it never really provided as much help as it seemed to. I'm trying to replace it with something with less overhead. All I can really make use of is the name of the last entered routine, the rest is just 'prettiness'.
A key logger may not even help with intermittent bugs, as there is obviously some underlying state of tab conditions, how they were opened, how they were switched between, etc. that exposes the error. Walking through a key log without that state knowledge would be fruitless as I'd be reviewing code which obviously works 99.999% of the time. The odds of 'seeing' a bug are basically nil.
This is more of a keep your eyes open for some pattern of activity that triggers the problem.
George
|
|