| Time |
Nick |
Message |
| 08:38 |
|
mmorgan joined #evergreen |
| 09:25 |
Bmagic |
eeevil++ miker++ goood++ |
| 09:25 |
Bmagic |
too many o's? |
| 09:26 |
goood |
just the right amount |
| 09:26 |
Bmagic |
perfect |
| 09:26 |
Bmagic |
just saw your IRC stuff above. Thanks for looking at that! I saw your reply on the bug yesterday too. I'll see if I can take that patch and make the error go away |
| 09:27 |
Bmagic |
The two users are system level. I even bumped it up to consortium level to hammer it, still no dice |
| 09:27 |
Bmagic |
*CONTAINER* AND *BUCKET* perms |
| 09:28 |
Bmagic |
you know how we do everything until something works. After an hour, and promoting the one user to Global Administrator and still got the bug, I reported it in LP |
| 09:29 |
csharp_ |
@ana TOO MANY MIKES |
| 09:29 |
pinesol |
csharp_: Error: I tried to send you an empty message. |
| 09:30 |
csharp_ |
pinesol: child please |
| 09:30 |
pinesol |
csharp_: I am the coolest bot. |
| 09:30 |
Bmagic |
lol |
| 09:34 |
|
Dyrcona joined #evergreen |
| 09:34 |
jeff |
brb, adding --23daves and --oops-all-mikes sample data load options. |
| 09:34 |
pinesol |
https://lp-oops.canonical.com/oops.py/?oopsid=all |
| 09:40 |
Bmagic |
jeff++ |
| 09:51 |
Dyrcona |
brb: Kernel update. :) |
| 10:03 |
|
mmorgan1 joined #evergreen |
| 10:14 |
|
Dyrcona joined #evergreen |
| 10:38 |
Bmagic |
well goood: your untested patch is now tested. The interface bugs without the patch, and doesn't bug with the patch, so I'd say it's working |
| 10:39 |
Bmagic |
and owning_lib is still null in the database |
| 10:48 |
|
Rogan joined #evergreen |
| 11:00 |
* Dyrcona |
cherry-picks the Angular 21 branch into a rel_3_17_beta test branch and resolves conflicts. Think I'll make a file with just the Angular 21 commits for easy cherry-picking later. |
| 11:05 |
Dyrcona |
Ooh. Lint issues. |
| 11:07 |
Dyrcona |
One of them was my fault during conflict resolution. |
| 11:10 |
goood |
Bmagic: sweet, thanks for testing. I really hope we can push as much down into the IDL and pcrud over time. speed is definitely better, but it's also going to be more secure. if there's one place to check /data access/ permissions, there's less chance of leaking. of course, action-restricting perms will always be needed, but hopefully exceptions to the access perms that action perms need to allow will be a vanishing small set. fwiw, the perl middle |
| 11:10 |
goood |
-layer container logic is exactly the sort of outside-direct-permission stuff I want to get rid of because of the leakiness (and, this LP and patch shows we should be able to, I think!) |
| 11:11 |
goood |
*as much as possible |
| 11:13 |
Dyrcona |
hm... Think I'll add a db upgrade script to the branch and then rebase to put it before the Angular 21 commits. |
| 11:27 |
Dyrcona |
larning'++ |
| 11:27 |
Dyrcona |
grr.. typos-- |
| 11:28 |
|
jihpringle joined #evergreen |
| 11:28 |
Dyrcona |
larnin'++ |
| 11:36 |
Bmagic |
goood: that all makes sense. I was thinking that we had some bugs lurking around performance where the root was doing too much* with pcrud and not enough with custom-made perl middle layer handling for performance |
| 11:44 |
goood |
I'm not going to say that could never happen -- there certainly could be ways to write individual, specific queries that would be faster than anything pcrud natively supports, and if that's a user-visible performance dealbreaker then we should do it -- but it's hard to imaging that in your general CRUD patterns. mostly because pcrud sits directly on the db, but perl (other than storage) has to go through another (there's a couple extra round trips), |
| 11:44 |
goood |
and the in-query permission testing means we never even have to see (and test visibility of) data that won't be returned |
| 11:46 |
goood |
that comes at the cost of complexity in IDL design (for most devs adding classes) or pcrud query construction (for the db-interface devs) |
| 11:47 |
goood |
but it means that we only have to think about permission complexity in one place (for access perms, not action perms, obv), which IMO is an overall benefit -- smaller problem space, less chance for a proliferation of "ways" |
| 11:53 |
|
adam_reid joined #evergreen |
| 12:25 |
|
Christineb joined #evergreen |
| 13:03 |
Dyrcona |
TIL about fstrim.timer |
| 14:26 |
|
Dyrcona joined #evergreen |
| 14:31 |
|
Dyrcona joined #evergreen |
| 16:28 |
Dyrcona |
Hey, great. I get different conflicts while cherry-picking the angular commits from rel_3_17 onto main.... Should be the same right now, no? |
| 16:30 |
Dyrcona |
Rather. It's looking like the same conflict again even though it should be resolved in the 3.17-based branch. |
| 16:50 |
Dyrcona |
scriptin'++ |
| 17:04 |
|
mmorgan1 left #evergreen |
