What are the rest 74% not covered by output?

[vimmerru]

Hi,

If just aggregate output from README.md it will be 26,2% in total only. What are the rest 74% not covered by output?

See attached screenshot

[RazrFalcon]

cargo-bloat parser only the .text section. There are many more sections, which are more or less optional/irrelevant. Here is a list of sections for ELF.

As for 26.2%, the last row is total. So 13.3% is a .text section size in the file. As mentioned in the Name column.

[vimmerru]

@RazrFalcon Yea... understand. Thanks for the explanation, but seems size analysis based on "text" section only doesn't help a lot in optimization as there is very limited correlation between total size and text section size.

[RazrFalcon]

All other sections are optional. Only the .text section contains your code. You can run strip on Linux and similar commands on other OS'es and most of the sections will be removed.

On the other hand, cargo-bloat indeed doesn't support static variables yet. But it should not affect the results in most cases.

PS: by the way, the original version of cargo-bloat didn't have the File column at all. I've added it only because someone asked for it. So cargo-bloat is designed just for "code size", not "binary size" analysis in mind.

[vimmerru]

In my specific case i got the following after strip. Text is 55%

Maybe other sections size correlates to text section size too.

[RazrFalcon]

350KiB for .rodata is a lot... And as mentioned, cargo-bloat doesn't supports constant/static data yet. This is basically #10

[vimmerru]

This is basically #10

I agree. Feel free to close the issue. I got the answers. Maybe a bit of explanation in readme will help. Thanks!