| Time |
Nick |
Message |
| 01:06 |
|
briank joined #evergreen |
| 01:06 |
|
jweston joined #evergreen |
| 01:06 |
|
eeevil joined #evergreen |
| 01:06 |
|
jmurray-isl joined #evergreen |
| 06:52 |
|
berick joined #evergreen |
| 07:16 |
|
collum joined #evergreen |
| 07:31 |
|
kworstell-isl joined #evergreen |
| 07:44 |
|
redavis joined #evergreen |
| 07:58 |
|
BDorsey joined #evergreen |
| 08:09 |
pinesol |
News from commits: Docs LP 2073411 Updating screens for SMS Messaging Page <https://git.evergreen-ils.org/?p=Evergreen.git;a=commitdiff;h=bfd4cfe8169f925bc7cb9da759218021c1a6a7e9> |
| 08:14 |
|
cbrown joined #evergreen |
| 08:35 |
|
ian1 joined #evergreen |
| 08:46 |
|
ian1 joined #evergreen |
| 08:47 |
|
redavis joined #evergreen |
| 08:49 |
|
dguarrac joined #evergreen |
| 09:09 |
pinesol |
News from commits: Docs LP2073412 Updating Circulation Limit Sets Page <https://git.evergreen-ils.org/?p=Evergreen.git;a=commitdiff;h=a8a7693407791cb4ec69a979bdf16165e492a98a> |
| 09:39 |
pinesol |
News from commits: Docs LP 2073415 Updating recent staff searches page <https://git.evergreen-ils.org/?p=Evergreen.git;a=commitdiff;h=a4c607801461344b9ac04be64d4a313e31d31410> |
| 09:46 |
Bmagic |
I know that Encode.pm can't be used on a multithreaded perl program but for the life of me I can't find the documentation on the issue. Anyone have an article that I can look at? |
| 09:51 |
|
Dyrcona joined #evergreen |
| 10:09 |
|
BDorsey_ joined #evergreen |
| 10:09 |
|
cbrown_ joined #evergreen |
| 10:55 |
Dyrcona |
customization-- |
| 10:57 |
|
kworstell_isl joined #evergreen |
| 10:57 |
Dyrcona |
That 1379 db upgrade is a tricky one. If you're upgrade goes from before 3.10 to after 3.10, you have to remember to include the 1379.function.user_circ_test.sql from 3.10. |
| 10:59 |
* Dyrcona |
plays with a branch to upgrade from 3.7.4(!) to 3.13.3(!).. Yeah, the later ain't out, yet, but that's what I'm calling it for the sake of versioning. |
| 11:08 |
Dyrcona |
Whoa! It's a 2.2MB upgrade file, and I'm sure it needs some editing.... |
| 11:25 |
Dyrcona |
Huh. Updates to Pg 10, but I thought it was EOL.... |
| 11:26 |
Bmagic |
Dyrcona: http://irc.evergreen-ils.org/evergreen/2024-08-02 do you know where I can find the article on the matter? |
| 11:29 |
|
BDorsey__ joined #evergreen |
| 11:29 |
|
cbrown joined #evergreen |
| 11:31 |
Dyrcona |
Bmagic: Well, one way to find out is to simply try it, and watch the script die with an error message. :) |
| 11:33 |
Bmagic |
I didn't think it would generate an error |
| 11:33 |
Bmagic |
I thoguht it was thought of as "unsafe" |
| 11:35 |
Dyrcona |
Perl 5 threads are garbage. It's a terrible implementation and no one uses them for anything serious. It's actually one of the main reasons Rakudo was invented, to fix threads. |
| 11:36 |
Dyrcona |
Ask anyone involved in Perl 5 development, and they will agree with me. ;) |
| 11:36 |
Dyrcona |
If a module in Perl is not thread safe, then it cannot be used in a threaded program There is no safe way to do that. |
| 11:37 |
Bmagic |
right, I'm on board with that. It's the Encode.pm module that I understood to be the deal breaker |
| 11:37 |
Dyrcona |
Yeah, Encode is not thread safe, though I don't see it mentioned in the documentation today. |
| 11:37 |
Dyrcona |
That may have changed, but I doubt it. |
| 11:38 |
Dyrcona |
Let me check, I might still have code hanging around where I tried to make it work... |
| 11:38 |
Bmagic |
scottangel is playing with threads using ParallelExecutor and I mentioned how we can't use Encode.pm with threads. And that led me to ask this question, because I can't find anything online explaining why |
| 11:39 |
Dyrcona |
Nah, doesn't look like I do. |
| 11:40 |
Dyrcona |
Yeah, I'm not finding anything either, so try it and see. If it works, great. If not, your script blows up. |
| 11:40 |
Stompro |
I also went down the threads+encode.pm road with marc_export.pm a few months ago.... Dyrcona tried to warn me also. |
| 11:40 |
Dyrcona |
Bascially, Perl 5 threads should not be used, ever. |
| 11:41 |
Dyrcona |
If you really want threads, use a different language, almost any other language, really. |
| 11:42 |
Dyrcona |
Java, Rust, modern C++ (i.e. C++11 or later), Python, all have much better threading than Perl 5. You could even try Raku if you want to see what Perl 6 turned into. |
| 11:43 |
Dyrcona |
I'd rather use POSIX threads in C than try to get something working with threads in Perl 5. |
| 11:43 |
Bmagic |
ParallelExecutor sounds like it's skirting the issue by using system() commands to spawn threads |
| 11:44 |
Dyrcona |
Bmagic: Wrong! Those are processes, not threads.... In that case you should be fine. |
| 11:44 |
Dyrcona |
Let's keep our terminology clear. Otherwise, it leads to confusion. |
| 11:45 |
Stompro |
Dyrcona, I just heard that in your voice, did you just say it out loud also? |
| 11:45 |
Dyrcona |
BTW, here's an "official" answer about Perl 5 threads: https://github.com/Perl/perl5/issues/14691 |
| 11:45 |
Bmagic |
fair enough. It accomplishes our goal of running more than one execution at the same time though |
| 11:46 |
Dyrcona |
Stompro: No, I didn't say it out load, but I did think it rather strenuously. |
| 11:46 |
Dyrcona |
Bmagic: It does, but processes and threads are very different in terms of architecture, especially since Linux no longer implements threads on top of processes. |
| 11:47 |
Bmagic |
My statements are framed as questions FYI. I'm exploring it with you, trying to figure out if it's safe or how to better guide scottangel |
| 11:48 |
Dyrcona |
I don't know anything about ParallelExecutor. If it uses processes, then you're good. If it uses Perl 5 threads, then you're done. My advice is pretty much always, "Try it and see." |
| 11:52 |
Dyrcona |
Bmagic: What module are you using? Parallel::TaskExecutor? If so, that looks OK. |
| 11:55 |
* Dyrcona |
steps out to grab a bite to eat. |
| 11:58 |
Dyrcona |
If you look at the source code of the module, and it does not say 'use threads;' then you are OK to go. |
| 11:59 |
|
jihpringle joined #evergreen |
| 12:04 |
Bmagic |
Parallel::ForkManager |
| 12:04 |
Bmagic |
gotcha |
| 12:13 |
berick |
Bmagic: apropos of nothing, this week I got Rust EG service workers running in k8s, one worker per pod. haven't setup scaling yet, though. Mostly just doing it to learn. |
| 12:16 |
Dyrcona |
Bmagic: I think Parallel::ForkManager is OK, too, but I haven't looked at it. |
| 12:16 |
Dyrcona |
berick++ |
| 12:17 |
Dyrcona |
I'm still reimplementing some of my personal utilities in Rust. I do that for practice when learning a new language. |
| 12:29 |
|
collum joined #evergreen |
| 12:55 |
|
Christineb joined #evergreen |
| 13:24 |
|
jihpringle joined #evergreen |
| 13:29 |
jonadab |
Makes sense. I usually start by re-implementing tetris. |
| 14:05 |
Dyrcona |
Oh! Got my custom db upgrade script to work with just a minor adjustment or two. Our data requires a CASCADE on one drop type in 1307, and the 1312 upgrade has to be moved to the end. |
| 14:15 |
Stompro |
Claiming 1424 , gonna give this a try. |
| 14:19 |
Stompro |
Dyrcona, does the claiming work any differently for something that isn't being backported to older releases... just trying to imagine what happens if 1425 is for a feature that is backported to 3.13, so then a 3.13 system has 1425 but not 1424? |
| 14:21 |
Stompro |
Or maybe I'm jumping the gun, does setting up the sql upgrade index number happen when something moves from master -> release branch? |
| 14:22 |
Stompro |
Looking at https://wiki.evergreen-ils.org/doku.php?id=dev:committing&s[]=claiming |
| 14:23 |
berick |
Stompro: the index number is set just before merging to main |
| 14:23 |
berick |
the process is the same regardless of whether it's backported |
| 14:24 |
berick |
release branches sometimes have gaps and that's OK |
| 14:24 |
Stompro |
berick++ thanks. |
| 14:30 |
Dyrcona |
yeah, and while I was not paying attention to IRC, I reviewed our config.upgrade_log and discovered that we're missing an upgrade from way back in the 700s... 0705. We don't have the permission it added in the database. |
| 14:54 |
Stompro |
Dyrcona, I have my branch ready to go, rebased to main, sql upgrade stamped. Can I push that to main... or is it better to cherry-pick on top of main and then push? |
| 14:55 |
Stompro |
Branch at https://git.evergreen-ils.org/?p=working/Evergreen.git;a=shortlog;h=refs/heads/user/stompro/lp1786524_patron_loader-3 |
| 14:57 |
Stompro |
I edited all the commit messages so all the LP tags match... makes my OCD happy. |
| 15:00 |
* Stompro |
needs to go drop off a UPS at fedex, back in 15 minutes. |
| 15:00 |
Dyrcona |
Stompro: Ok. I'll wait. |
| 15:01 |
Dyrcona |
What's up with Rogan not signing his commits? |
| 15:01 |
Dyrcona |
That alone should stop this from going in. |
| 15:02 |
Dyrcona |
I know a project *monero cough* where cryptographic signatures are required.... |
| 15:25 |
|
jihpringle joined #evergreen |
| 15:30 |
Stompro |
Back, sorry. Couldn't even drop off the UPS, no one ever came to the walgreens photo counter. |
| 15:31 |
Stompro |
About the sign off, does that need to be dealt with before bug 1786524 gets committed? |
| 15:32 |
pinesol |
Launchpad bug 1786524 in Evergreen "Add a support script for importing patrons" [Wishlist,Confirmed] https://launchpad.net/bugs/1786524 - Assigned to Josh Stompro (u-launchpad-stompro-org) |
| 15:38 |
Dyrcona |
Stompro: Yes, because we don't allow force pushes to main. |
| 15:38 |
Dyrcona |
They get automatically rejected. |
| 15:40 |
Stompro |
Dyrcona, sorry, was that in response to my "How to commit a branch" or "Rogan's signoff"? |
| 15:45 |
Dyrcona |
It was in response to Rogan needing to sign off before it's committed to main. |
| 15:46 |
Dyrcona |
Stompro ^^ |
| 15:49 |
Stompro |
Gotcha... I'll try and watch for that in the future... |
| 15:49 |
Dyrcona |
Stompro: The only way to add the signed-off-by to the commit now is to rebase the branch. If that is done after it goes into main, then it would have to be force-pushed and our configuration will reject that. |
| 15:50 |
Dyrcona |
Above was as much for the logs as for Stompro. |
| 15:50 |
Stompro |
Thanks for the extra explanation Dyrcona. |
| 15:54 |
Stompro |
I'm guessing it would be ok for me to add it if he just comments on the bug report? |
| 16:02 |
Dyrcona |
Yeah. That would be OK. |
| 17:06 |
|
book` joined #evergreen |
| 22:58 |
|
scottangel joined #evergreen |
| 22:59 |
|
Jaysal joined #evergreen |
| 22:59 |
|
dluch joined #evergreen |
