search

geiger-rs / cargo-geiger

Detects usage of unsafe Rust in a Rust crate and its dependencies.
https://crates.io/crates/cargo-geiger
1.41k stars 67 forks source link

A broken output format on windows? #252

Open zhiburt opened 2 years ago

zhiburt commented 2 years ago

Hi there,

Just noticed that a table seems to be a little bit tricked. I do not certain if this is a windows issue.

You can reproduce it by.

git clone https://github.com/zhiburt/expectrl
cd expectrl
cargo geiger
Functions  Expressions  Impls  Traits  Methods  Dependency

855/3420   6393/25572   1860/7440 0/0     50/200   !  expectrl 0.2.0
0/0        219/219      0/0    0/0     0/0      !  ├── conpty 0.3.0
1223/17405 18323/1389609 823/48930 3/3     114/77943  !  │   └── windows 0.29.0
0/0        0/0          0/0    0/0     0/0      ?  │       └── windows_x86_64_msvc 0.29.0
0/0        34/34        1/2    0/0     2/2      !  └── regex 1.5.4
19/19      678/678      0/0    0/0     22/22    !      ├── aho-corasick 0.7.18
36/37      2067/2140    0/0    0/0     16/16    !      │   └── memchr 2.4.1
0/19       0/319        0/0    0/0     0/30     ?      │       └── libc 0.2.101
36/37      2067/2140    0/0    0/0     16/16    !      ├── memchr 2.4.1
0/0        0/0          0/0    0/0     0/0      :)     └── regex-syntax 0.6.25
pinkforest commented 2 years ago

Thanks for bringing it up :purple_heart:

The column width maximums we use overflow here -

We'll need to improve output formatting to ensure we don't overflow the max column widths in order to keep reasonable max width

I am thinking that instead of for example 1389609 (LOL!) we should just have 1.4M with k/M etc. units shortened to ensure column width guarantees to prevent overflow - given the counter would be 1.4M to begin with.

Another thing we need to do is to count the unsafe's correctly (not sure yet whether this count is valid but makes good test case) for the windows crate - thanks for bringing another test crate subject we can run CI regressions against on next major.

:unicorn: