"Benchmarks for artifact sizes" confuses me

[chengr4]

According to @Kobzol 's blog, The main function of helloworld-tiny and ripgrep-13.0.0-tiny are to monitor the size of Rust executable

Refer to blog: There are also two secondary benchmarks (helloworld-tiny and ripgrep-13.0.0-tiny), which exist solely to monitor the size of a “minimal-sized” Rust executable, which is compiled with compiler flags that should favour small executable size.

However, on the graphs page, the title states 'Benchmarks for artifact sizes', but the metric displayed is "CPU instructions", which confuses me.

I believe it should present the value of size:xxxxxx.

Solutions I come up with

1. (Simple) Write a note below the title explaining that it only makes sense when the kind is size:xxxxxx.
2. We should separate the APIs and let them have their own behavior

---

related to : #1817

[Kobzol]

Hi, the whole graphs page is displayed with a single metric, so if you choose instructions, you'll see instruction counts for all the benchmarks.

While the binary size of these tiny benchmarks is indeed the main motivation for why we have added them, their compilation time ia actually not uninteresting! They use different compiler flags, and LTO, so it is also interesting to look at how does their compilation time change.

[chengr4]

@Kobzol Thanks for explanation.
Then, I think none of my solutions is feasible for this issue.
But I still think the title "Benchmarks for artifact sizes" confuses the user (damage UX).
What about we delete this category?

[Kobzol]

Feel free to think of a better label. For example, Benchmarks optimized for binary size or something like that.