I have looked at the code. There is only one method for reading and one for writing and they both close the file when they have done so thatâs wouldnât ât seem to be the problem. It still doesnât answer why it works on my M1 MacBookAir and not on two Windows boxes :-(
On 20 Apr 2021, at 16:13, Sven Van Caekenberghe <sven@stfx.eu> wrote:
David,
On 20 Apr 2021, at 17:01, David Pennington <david@totallyobjects.com> wrote:
I have finally given up on trying to sort this out. Under both Windows Server 2012 and Windows 10 my Pharo image just exits leaving nothing, not even a crash file. Hence, I have fixed my MacBookAir so that it stays alive with the lid down and I am now using that as my Seaside server. It has been up for the last 4 days so it seems that this is a Windows only problem and (hopefully) not my code.
In passing, I am a high-level programmer in Pharo in that I do not use any special calls or processes.
Davids Totally `Objects
I believe you are using some form of persistency that you built yourself, with data written in STON to files. Assuming you store each object in its own file, you are then also doing a lot of IO. Are you sure that you are properly closing your file streams, in all circumstances ?
It is possibly that you run out of IO resources.
Also, garbage collection on a server image (without a UI) is sometimes a bit strange, it can take a while to kick in.
I am just guessing here.
It is still not good that the image/vm quits without any notice, that is for sure.
Sven