Incorrect binary size error on trybot?

[Khushal Sagar]

Hello fuchsia folks,

My change is triggering a binary size increase error on the fuchsia-binary-size. It's adding a new blink feature with mojom IPC plumbing via content. No new libraries being imported or any autogen files being added. It's unclear to me where this increase could be coming from. Is it possible that the error is incorrect?

Thanks.

Khushal

[David Dorwin]

Hi Khushal,

That is surprising for this CL. PS8 failed twice then PS9 and later succeeded. There's nothing obvious in the diff between those two patchsets either. Maybe the extra Mojo calls were big somehow.

+rohp...@chromium.org, any ideas?

--
You received this message because you are subscribed to the Google Groups "fuchsia-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to fuchsia-dev...@chromium.org.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/fuchsia-dev/CAMLuWUxMtQynt%2B6dLkof6NH6OWVQmXRBE7hesbCCZNxy35oG0Q%40mail.gmail.com.
For more options, visit https://groups.google.com/a/chromium.org/d/optout.

[Rohan Pavone]

Unfortunately, there is not a detailed breakdown without manually using bloaty to examine the results (documented). However, mojom changes can cause unexpectedly large increase. From the android-binary-size bot, we can see there is a 6k+ increase. While WebEngine is not built the same as Android, this helps us identify potential sources of increase. 

From the following screenshot, you can see that there was an increase from introducing this change). You can see the detailed view here. This tells me that this is a real increase, albeit small.

As for fuchsia-binary-size (which does not have this nice view yet), patchset 9 did cause an increase in uncompressed growth, but the compression algorithm somehow managed not to include any growth, so it did not fail (there was a 12k increase).

Patchset 8, on the other hand, had a 16k increase in compressed size, thus failing.

Between these two changes, Patchset 8 had more growth in $NATIVE (not sure how that is determined would need to talk to Android about what this means), vs Patchset 9 where $NATIVE actually shrunk, and //content had small overall growth. //content also grew by less due to moving code into blink.

Not sure what goes into $NATIVE, but that probably impacted both. Sorry for not being able to provide a better answer.

On Wednesday, November 2, 2022 at 6:02:47 PM UTC-7 David Dorwin wrote:

Hi Khushal,

That is surprising for this CL. PS8 failed twice then PS9 and later succeeded. There's nothing obvious in the diff between those two patchsets either. Maybe the extra Mojo calls were big somehow.

+rohp...@chromium.org, any ideas?

On Tue, Nov 1, 2022 at 8:57 PM Khushal Sagar <khusha...@chromium.org> wrote:

Hello fuchsia folks,

My change is triggering a binary size increase error on the fuchsia-binary-size. It's adding a new blink feature with mojom IPC plumbing via content. No new libraries being imported or any autogen files being added. It's unclear to me where this increase could be coming from. Is it possible that the error is incorrect?

Thanks.

Khushal

--
You received this message because you are subscribed to the Google Groups "fuchsia-dev" group.

To unsubscribe from this group and stop receiving emails from it, send an email to fuchsia-dev+unsubscribe@chromium.org.