hey gents, I'm seeing different behaviour pertaini...
# random
g
hey gents, I'm seeing different behaviour pertaining to a
process.start()
call from when I run the target under intelliJ, and under an IntelliJ ant target than when I run it from install4j. This is on windows 7. Interestingly the behavior difference disappears on our (newer) windows 10 machines. Anybody here got some thoughts about what the difference might be?
Oook, so the problem was
javaw
vs
java
. Windows apparently has some concept of a console window, where processes can either have a console window or they can not have a console window (even if they spawn several later in their life cycles). Abaqus, being an ancient piece of software, was not willing to tolerate being `Process.start()`'d by a young-upstart-non-console'd process, which is what our software is. There must be come old API on the windows process object. Infuriating.
Also the exact behaviour of this flag/property appears to have changed from Windows 7 to Windows 10
d
Ouch. Did they actually document the change anywhere?
g
probably? but I'm not even an MDSN subscriber, so there was no way I was in a channel to get news of that change. It could also be the software doing an
if(WIN_7) doA() else if (WIN_10) doB()
, I'll never know
but yeah it was probably 30-40 man hours to solve this stupid flag.
d
This did seem like hard won knowledge.