Lin Jen-Shin (9a134fb2) at 16 Mar 18:18
Shorten the SHA1 so that /logs has shorter width for mobile
Lin Jen-Shin (6485e15f) at 16 Mar 16:56
Improve mobile/phone view by better utilizing spaces
... and 1 more commit
Lin Jen-Shin (772e04b3) at 16 Mar 13:48
Update event data
Lin Jen-Shin (2656d37c) at 11 Mar 11:40
Update event data
This is updated in f1bd4ef9
Rarity is now visible for owned cats — the cell shows the rarity color on the left and azure on the right, instead of azure overriding the rarity color entirely. Note: CI may fail on my fork due to GitLab's unverified identity requirement, not a code issue. All tests pass locally.
Lin Jen-Shin (f1bd4ef9) at 09 Mar 12:33
Wrapping up color coding for owned cats
... and 6 more commits
Rarity is now visible for owned cats — the cell shows the rarity color on the left and azure on the right, instead of azure overriding the rarity color entirely. Note: CI may fail on my fork due to GitLab's unverified identity requirement, not a code issue. All tests pass locally.
@asHOH Sorry for the delay and thanks a lot, this looks good!
It's actually slightly different than what it was before, but it's also looking better. On Chrome it looks like it even fixed a small 1px gap. Well done!
No problem. Hope you had a great vacation.
Hmm. From my experience, division by 0 would usually crash the program (especially with the same compiler and compile option, clang -O2 in this case), and I forgot it's probably also dereferencing a pointer from malloc(0). From my experience that could indeed lead to getting random numbers, and then it might break the conditions for the loop. Maybe that's how it's running indefinitely on Mac.
While we have jokes on "undefined behaviour" to launch a nuke, or Principle of Explosion to derive someone's phone number, I would still expect some explanation since a lot of that were done by practical reasons.
Anyway, back to the topic. I checked the log and this case seems to still happen a lot.
Sometimes I do see browser can somehow reset the banner under some conditions (like restoring from cache, or back page or something like that?), and that's probably why some did pick platinum banner for seeking rare cats...
It happens often enough that I think we should still address this.
I think ideally the seeker shouldn't crash in any way. However, the input is indeed also invalid, so I think it does make sense to check it before passing to the seeker as well.
This one is less easy to check, but I think we can still do a quick sanity check to ensure that slots are not going out of bound.
It would be the best to prevent the browsers from resetting banners like this, but I am not sure how without knowing how exactly why the browser is doing that.
In summary I think we have 3 things here:
We should address all of them, but I am ok if we just check the input to address this for now.
Lin Jen-Shin (52bc84e1) at 09 Mar 09:54
Update BCJP app data
Lin Jen-Shin (110a61e4) at 09 Mar 09:46
Update app data and add talent counter-surge
There's another crash similar to this, but other than that, it seems to be working fine now.
Lin Jen-Shin (c9412aad) at 04 Mar 18:11
Set OOMPolicy=continue for Varnish