Platform<=>version conflict or what?

[Jorma Paavola]

Hi Phil,


>> This works fine on my Win2k ! - It'll create Runtime-Temp, where the 
>> TestApp is running from, when started and erases Temp-file automatically
>> when exited - this feature I like !!
> 
> OK,  I see better what you are trying to do now,  building a single
> executable file.  I haven't tried this,  but the questions I'd have
> about it would be:
> 
> - Have you verified that the files have been properly extracted
> from the single executable into the temporary directory?  Do they
> have the same MD5SUM's as they did before they were placed into the
> single file executable?

Unfortunately extracted files cannot be seen from Temp-folder, because 
it will be automatically erased when exited (or ErrorMsg given) !

> 
> - Does it make any difference if compression is turned off?
> 
> - Maybe it's an order of events thing?  (maybe the .dlls have to
> already be present before the .exe starts running?? I ask this not
> knowing if it extracts them all,  or if they are extracted "on demand"
> as requested,  or the order in which they are extracted...)

Principally NSIS should take care of prompt order of events !

> 
>> What is WINO386.mod ???
> I'm not sure but this may be something called "WinOldApp" that I
> sometimes see running in the task manager list.

You are right: that was related to WinOldApp !

> 
>> 4)Filemon
>> - Can you collect the whole trace with it ?
> I've never run into a case where I could not see what I wanted to see
> with Filemon.  I usually let things run to the point where the event
> has happened, then stop the capture in Filemon,  and look back at what
> happened.

I tested MyApp on Win98 with "FileMon" to see what happens - this tool 
really traces everything !
- it found two 'unsuccesful' (missing?) modules: DOSAPP.INI and CONAGENT.PIF
- are these important ?
(- also sometimes the order/function was strange: f.ex. trials to CREATE 
already existing files like C:/WINDOWS, C:/WINDOWS/TEMP etc.)
(- also sometimes attributes were tested first before the file was 
created ; thereafter everething worked fluently again ...)
- it looks quite difficult to localize the event(trigging/failure 
point)from long list totalling abt. 6.000 events (from start to Errormsg) !
Any ideas for that ?


Rgds,

Jorma