Pull request: Fix for "Cloning HG repo to Git creates "uncheck-out-able" repositories". And upgrade for filenames transcoding

[IngeniousOne]

Hi, everybody!

I have two suggestion for you.

• First: A bug-fix for a problem on Windows systems:

   Cloning HG repo to Git creates "uncheck-out-able" repositories 
  

because

crushes paths (e.g. /bin/foo-file.py => \bin\foo-file.py), making them unknown for Git repo (git uses "/ " for storing file paths instead of "\") on check-out. In reply to "checkout" command it generates error message:

first commit fixes that.

• Second:
  A solution for cases when HG repository contains files with names in non Utf-8 encoding and containing culture-related symbols (e.g. cp1251 "Тестовый файл.txt").
  In this cases, cloning a HG repository to git with git-remote-hg, produces unreadable filenames in git
  
  Solution bidirectionally solves this problem by transcoding filenames on-the-fly. All you need is to add to .git/config or to a global one options:
  • "remote-hg.transcode-filenames" - boolean key option to enable|disable transcoding functional;
    • remote-hg.hg_filenames_enc - required string - HG repo filenames encoding;
    • remote-hg.git_filenames_enc - optional string(default: utf-8) - Git repo filenames encoding.

[benbrummer]

Can you add this PR to the mnauw fork, too. Almost all Distributions (Ubuntu, Fedora, Arch Linux...) are useing this. In fact this error is already fixed there, but your solution is in my optinion the better one.

[IngeniousOne]

Hi. Ok, all add my pull request in mnauw's fork too. пн, 18 нояб. 2019 г., 14:59 benbrummer <[email protected]>:

…

Can you add this PR to the mnauw fork, too. Almost all Distributions (Ubuntu, Fedora, Arch Linux...) are useing this. In fact this error is already fixed there, but your solution is in my optinion the better one. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#79?email_source=notifications&email_token=AMXPO4QPREU77R2I7I5EETLQUJRRZA5CNFSM4IH6S5UKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEEJ4EMI#issuecomment-554943025>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AMXPO4QZDNNATFFPPL62E5LQUJRRZANCNFSM4IH6S5UA> .

[felipec]

Wouldn't it be better to have a per-repository configuration that specifies the encoding of the Mercurial repository? You wouldn't want to transcode as cp1251 all repositories.

Also, in Git all repositories are assumed to be utf-8, so there's no need to specify that.