Negative binary size increase after doing changes that shouldn't affect the binary size

[Moritz Firsching]

We are currently trying to understand why the android binary size try bot reports a decrease in binary size for two commits which shouldn't affect the android binary size 

https://chromium-review.googlesource.com/c/chromium/src/+/2745343

and

https://chromium-review.googlesource.com/c/chromium/src/+/2749318/3

For both the try bot reports changes of around 45k:

Android Binary Size (-44,758 bytes)
Base Module Size (-44,758 bytes)

In the first of those just some libraries are added without being compiled and the second one has a build flag which is disabled on android builds. If it was enabled we would be expecting a size increase not a decrease. 

What could be the cause of this issue? Are we somehow not using the try bot correctly?

[Andrew Grieve]

The second one depends on the first, so the question is just why does the first one show an improvement.

This size breakdown shows that the decrease is in IDR_ABOUT_UI_CREDITS_HTML, which is a brotli-compressed blob of third_party LICENSE files.

Your CL is adding new LICENSES, so my best guess is that it's shuffling the file in a quite advantageous way for brotli. It could also be some result of stale license files being cleaned up by the bots, but given that it repro'ed in both of your commits, I think that's less likely.

Will be interesting to see if the result sticks when you commit (we should see it on perf bots).

--
You received this message because you are subscribed to the Google Groups "binary-size" group.
To unsubscribe from this group and stop receiving emails from it, send an email to binary-size...@chromium.org.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/binary-size/ff8c2b5a-fea6-4c7b-aaaf-39d9f5089b19n%40chromium.org.

[Moritz Firsching]

Thanks for the quick response, Andrew!

The explanation with brotli seems plausible.