Bugzilla

Do you mean this message?  or maybe I overlooked some other message?

> It is not recommended to compare a try build against a mozilla-central build, unless it is based on latest mozilla-central.

for me, it sounds like "do not compare try vs m-c if base revisions are far different", especially by the "unless it is based on latest mozilla-central" part that adds an exception to " the first sentence.
and, my understanding from it is "it's fine to compare try vs m-c if the try push is based/rebased onto latest m-c",
that's the situation where I hit the issue.

So, to my understanding, "comparing the same revision built on try push and m-c push" case isn't covered by the message,
and also it doesn't imply underlying performance characteristics difference between try vs m-c pushes, that's the actual issue the developer should be aware of, and the reasoning why "not recommended".

I'd like to reword it to something like:

"It is not recommended to compare a try build against a mozilla-central build.  the performance characteristics can be different between them.  Please consider pushing the mozilla-central revision also to try server and comparing between try builds."

Also, if the performance characteristics difference and the underlying binary difference are real, is there any valid scenario for comparing a try build vs a m-c build?
if there isn't any, disallowing the comparison will prevent similar trouble in the future.

Do you mean this message?  or maybe I overlooked some other message?

> It is not recommended to compare a try build against a mozilla-central build, unless it is based on latest mozilla-central.

for me, it sounds like "do not compare try vs m-c if base revisions are far different", especially by the "unless it is based on latest mozilla-central" part that adds an exception to the first sentence.
and, my understanding from it is "it's fine to compare try vs m-c if the try push is based/rebased onto latest m-c",
that's the situation where I hit the issue.

So, to my understanding, "comparing the same revision built on try push and m-c push" case isn't covered by the message,
and also it doesn't imply underlying performance characteristics difference between try vs m-c pushes, that's the actual issue the developer should be aware of, and the reasoning why "not recommended".

I'd like to reword it to something like:

"It is not recommended to compare a try build against a mozilla-central build.  the performance characteristics can be different between them.  Please consider pushing the mozilla-central revision also to try server and comparing between try builds."

Also, if the performance characteristics difference and the underlying binary difference are real, is there any valid scenario for comparing a try build vs a m-c build?
if there isn't any, disallowing the comparison will prevent similar trouble in the future.