Anyone testing this on the 15.1 beta channel? (Uncharacteristically of Apple, they were running simultaneous betas on 15.0 and 15.1). One imagines 15.1 shall be out soon.
Give me something to test and I can run a test. But so far I haven’t seen any problems with my mostly IMAP and SMTP connections.
Next week. Max Oct, 28.
Setup a mysql server on another computer on your local network.
Try to connect to it using the commands found in any example (db.connect). It will fail unless you db.connect, Thread.SleepCurrent(100), then db.connect again.
I have a dozen apps that use mysql and all of them fail on 15.0.1, whether compiled or run in the debugger.
It could be the local network part that’s the problem. Looks like Apple doesn’t want apps talking to each other locally. I found the same connection problems using MS Remote Desktop. Have to constantly toggle the app in Privacy>Local Network Access to get it to connect.
Haven’t tried 15.1 as I just went through all the trouble to roll back to 14.
I’d like to point out that since using 24r3 and 3.1, I’ve been the same startup issues with URLConnection on macOS 14 as well. The problem is definitely worse in 15, but definitely present in 14 for the first 2 or 3 seconds of app runtime.
So, you mean that URLConnection from 24r2 on Sequoia works ok, but from 24r3.x is flaky?
I’m saying that URLConnection from 24r2 works properly on Sonoma and from 24r3.1 it does not.
Anyway, that isolates the URLConnection problem to Xojo, not the OS. The 24r3.x seems compromised.
Also, there are reports of MySQL flaws too, but seems only on R3.1, not on R3.
Weird how small patches fixing few things introduces new bugs in Xojo. I was expecting R3.1 as a R3, but better, but it isn’t.
I’m not convinced. I’ve had LOTS of networking problems. I couldn’t even finish a large transfer from my NAS (in Finder) until I turned off my firewall. And my connection issues still happen in 2023r4.
Is there a case for this? (I looked and didn’t see one)
I think here:
Rick is referring to an issue I was having with in 2024r3.1 with MySQL.
I put in a case of the MySQL bug and it’s marked as fixed and scheduled for 2024r4:
I think no one created a sample and reported one.
Would be great if someone could make one.
The closest to the subject (URLConnection) was closed by the bot as “not replied in 2 weeks”, but it was about another weird thing, uncatchable exceptions.
https://tracker.xojo.com/xojoinc/xojo/-/issues/77586
@Rick_Araujo Agreed. (it would be great if someone who had a sense of the issue could make a case)
@Greg_O - seems like you might have an inkling of what’s going on? (especially given your deeper knowledge of the inner working of Xojo).
Do you think you have enough info to put together a decent case?
Wish I did. I think I could easily reproduce it though and that might be enough. I’m beginning to believe that signed apps are taking longer to finish launching than unsigned ones do though. That would be explain why it seems so intermittent to me. I sign most but not all things.
I haven’t done extensive testing, but I think I’m not seeing it on 15.1.
Unfortunately I am seeing it on my M1 under 15.1 for my recent development of APT. My solution was to introduce an artificial 3 second delay before the first socket is created.
That puts the problem on the Xojo 24r3.x field, not the OS.
If I’m not mistaken, one difference introduced by R3.x, maybe involved, were some sleep(n) inconsistencies. Like r3 having n = 999 acting like n = 1, and R3.1 “fixed” it, but not, just changed the behavior, and it maybe introduced extra problems in some other place now sleeping inconsistently.
https://tracker.xojo.com/xojoinc/xojo/-/issues/77567
https://tracker.xojo.com/xojoinc/xojo/-/issues/77691
A test case is necessary to avoid the problem (and maybe others correlated but not detected yet) slipping into the R4.
FYI I’m having similar problems and since it’s 2 years later I thought it better to start a new thread here. urlconnection-1009-errors-for-local-connections