Output is terrible on windows

[kjeremy]

This is what it looks like on rust-analyzer.

                                         �[1;37mProject: �[0mrust-analyzer
�[37m             `  :y.`yy`.y:  `�[0m            �[1;37mHEAD: �[0m0cce2bc (master, origin/master, origin/staging)
�[37m         -``MNsNMMNNNNMMNsNM``-�[0m          �[1;37mVersion: �[0mguide-2019-01
�[37m      ` -MMNMMMMNNm``NNNMMMMNMM- `�[0m       �[1;37mCreated: �[0m1 year, 10 months ago
�[37m     `NNNMMMdo:` `+md/  `:odMMMNNN`�[0m      �[1;37mLanguages: �[0mRust (96.38 %) TypeScript (3.57 %) JavaScript (0.04 %)
�[37m   -ssNMMNo.                .oNMMNss-�[0m               Shell (0.01 %)
�[37m   `mMM�[0m�[91mMMNmmmmmmmmmmmmmmmdy+`�[0m�[37m `sMMMm`�[0m    �[1;37mAuthors: �[0m40% Aleksey Kladov 2128
�[37m `mMMM�[0m�[91mMMMMMMMMMMMMMMMMMMMMMMN/�[0m�[37m  hMMMMm`�[0m  �[1;37m         �[0m25% bors[bot] 1313
�[37m -oMN-:Ny:�[0m�[91mmMMMMMm    oNMMMMMm�[0m�[37m oN::MMo-�[0m   �[1;37m         �[0m5% Florian Diebold 296
�[37m.yMMMhhh+ �[0m�[91mdMMMMMd:::::+mMMMMN/�[0m�[37m odyhMMMy.�[0m �[1;37mLast change: �[0m4 hours ago
�[37m-sNMMy    �[0m�[91mdMMMMMMMMMMMMMMMMs`�[0m�[37m    `yMMNs-�[0m �[1;37mRepo: �[0mhttps://github.com/rust-analyzer/rust-analyzer.git
�[37m-sNMMy    �[0m�[91mdMMMMMNyyyydMMMMMMy�[0m�[37m   .odMMNs-�[0m �[1;37mCommits: �[0m5227
�[37m.yMMMm   �[0m�[91mdMMMMMh     +MMMMMM+�[0m�[37m  sMMMMMy.�[0m  �[1;37mLines of code: �[0m61872
�[37m -oMMM�[0m�[91mMMMMMMMMMMMMMM+  mMMMMMMMMMM�[0m�[37mMMMo-�[0m  �[1;37mSize: �[0m18.96 MiB (889 files)
�[37m `mMMM�[0m�[91mMMMMMMMMMMMMMM+  :NMMMMMMMMM�[0m�[37mMMMm`�[0m  �[1;37mLicense: �[0mApache License 2.0
�[37m   `mMMMm               `-:o+:/mMMMm`�[0m
�[37m   -ssNMMMyomo            smohMMMNss-�[0m
�[37m     `NNNMs+mN/-`      `-/Nd/yMNNN`�[0m
�[37m      ` -MMNMMMMMNmmmmNMMMMMNMM- `�[0m
�[37m         -``MNsNMMNMMNMMNsNM``-�[0m
�[37m            `  :y.`yy`.y:  `�[0m

[spenserblack]

Looks like this might be fixed by updating colored to 1.8, which includes support for the Windows build target.

[kjeremy]

That did not fix it locally. I can run with NO_COLOR=1 in my environment and that at least gets readable output.

[spenserblack]

Which version of Windows is this? Just ran it on W10 (without updating crate) and it works. This is probably related to colored-rs/colored#59.

[kjeremy]

1809 Build 17763.805

[kjeremy]

The examples from the colored repo work

[kjeremy]

Even on tag v1.6.1 so it might not be a colored thing specifically.

[kjeremy]

Great that fixed it!

[o2sh]

@kjeremy can you give it another try? I've just pushed an update in the light of @spenserblack remarks.

[o2sh]

I'm confused, did my commit fixed it ?

[kjeremy]

Yes your commit fixed it :)

[spenserblack]

Great that fixed it!

🎉
Not sure why it worked on my W10 machine before the commit (older version, too) 🤔

[o2sh]

Did you run the .exe or just via cargo run ?

Anyway, I'm not sure my commit is the best way to handle this issue.
I didn't use control::set_virtual_terminal(bool)

========================

Thanks again for the help @spenserblack .

I added you as a Collaborator of the repo. You can merge pull requests when you feel like it's of enough quality and commit to the repo directly 👍

[o2sh]

The output is broken again 😢

[spenserblack]

😱
I didn't really have a chance to test it on Windows (since the colors worked before the fix for me for some reason, still not sure why).
Did I perhaps reverse the logic here when getting it to compile on Linux?

[kipters]

Did you test on both the old conhost.exe console and the new Windows Terminal? Maybe there's a difference there

[spenserblack]

Just pushed up a new commit which should fix the issue. Was able to replicate behavior on my Windows machine.