Timeline for How can I resubmit data to a SocketConnect object without creating a new socket?
Current License: CC BY-SA 4.0
14 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Aug 16, 2020 at 14:45 | comment | added | Shredderroy |
Thank you so much again! Yes, I was testing it by using Pause[60 * 15] in WL, and ensuring that when the programme resumed, the TCP server still returned the results. I need a long-running server, which I intend to call over the course of the generations of my evolutionary algorithm. The complete programme takes over eight hours to run, so I definitely need the server to accept requests after considerable delays.
|
|
| Aug 16, 2020 at 14:39 | history | edited | flinty | CC BY-SA 4.0 |
fix timeout issues - there is now no timeout at all on the socket
|
| Aug 16, 2020 at 14:37 | comment | added | flinty | @Shredderroy I've removed the timeout completely - it should stay up indefinitely until the connection is closed even with considerable delay in between sending. | |
| Aug 16, 2020 at 14:33 | history | edited | flinty | CC BY-SA 4.0 |
fix timeout issues - there is now no timeout at all on the socket
|
| Aug 16, 2020 at 14:00 | history | edited | flinty | CC BY-SA 4.0 |
timeout fixes
|
| Aug 16, 2020 at 13:48 | comment | added | flinty | @Shredderroy Ah, in that case there's still something wrong with my code then, because the time delay is arbitrary. If recv gets no data in the time frame then it raises an exception and this finishes off the connection. Have a look at this answer stackoverflow.com/a/16745561 . I'll see if I can modify the server code to fix this for arbitrary timeout. It also looks like serve_forever timeout anyway stackoverflow.com/a/27830693 so if you don't handle the whole request quickly enough and your second Map is issued too late, the connection breaks down. | |
| Aug 16, 2020 at 13:41 | comment | added | Shredderroy |
Thanks so much again. You know what made the difference? I was using time.sleep(0.05). Then I changed it to time.sleep(0.06), and now all the various maps get evaluated correctly.
|
|
| Aug 16, 2020 at 12:27 | comment | added | flinty |
@Shredderroy Strange - maybe there is still a race of some kind. It works for me if I do sock = SocketConnect[...] followed by many maps followed by Close[sock]; I even tried with a Do loop instead of a single Map: vals = Reap[Do[Sow[Map[getResFromPython[sock], {"arg1", "arg2", "arg3"}]],10]][[-1, 1]] and this produces 30 results in python and 10 lists of {"{\"res\": 1}", "{\"res\": 1}", "{\"res\": 1}"} in mathematica. Are you sure you cleared out all old definitions and copied the above code correctly for both server and Mathematica ?
|
|
| Aug 16, 2020 at 2:20 | comment | added | Shredderroy |
Another quick question. When I try to evaluate two Maps, the second one never gets evaluated. vals1 = Map[getResFromPython[sock], {"arg1", "arg2", "arg3"}]; vals2 = Map[getResFromPython[sock], {"arg11", "arg12", "arg13"}]; The first one gives me the correct results, but the second one never seems to reach the server,
|
|
| Aug 16, 2020 at 1:50 | vote | accept | Shredderroy | ||
| Aug 16, 2020 at 1:50 | comment | added | Shredderroy | You, sir, are a genius! Thank you for that detailed and painstaking response! | |
| Aug 15, 2020 at 23:37 | history | edited | flinty | CC BY-SA 4.0 |
edited body
|
| Aug 15, 2020 at 22:48 | history | edited | flinty | CC BY-SA 4.0 |
deleted 8 characters in body
|
| Aug 15, 2020 at 22:41 | history | answered | flinty | CC BY-SA 4.0 |