Compress on the fly #82
Conversation
This is a breaking change, since the format of the files we write has changed, and the process to open them has differed.
|
🤔 hrm tests are failing now, even though we are decoding. i'm not sure why. Any idea? |
|
@adnan-alhomssi: This seems like a good idea, and is related to your suggestion that we should be compressing the profiles and heapsnapshots before sending them back from ProfileEndpoints.jl. Can you have a look and see if you can help us figure out why the tests aren't passing? |
|
Request for review: @Drvi |
Codecov Report
@@ Coverage Diff @@
## main #82 +/- ##
==========================================
+ Coverage 96.59% 96.66% +0.06%
==========================================
Files 3 3
Lines 294 300 +6
==========================================
+ Hits 284 290 +6
Misses 10 10
📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more |
|
Will this break reading profiles from older PProf.jl versions? Typically with text files, one can sniff the gzip "magic number", but with ProtoBuf I think we need to check first if there are possible collisions with the encoded message. |
|
Thankfully, i don't think PProf.jl ever reads profiles, we just hand them over to So i think it's okay? What concern did you have in mind? |
|
Ah if we don't need to read it on Julia side, then it's fine:+1: sorry for the noise |
Fixes #81