Standardize algorithm for directory hashing #100
Conversation
|
cc @wolfv |
|
In pixi we often compute these hashes in a similar way. We also normalize line endings in text files because on windows git often normalizes these as well. |
Nice that you mentioned this because I think I just run into that very issue: https://github.com/conda/conda-build/actions/runs/11924575237/job/33235172020?pr=5277#step:11:424 |
cep-00??.md
Outdated
|
|
||
| ## Specification | ||
|
|
||
| Given a directory, recursively scan all its contents (without following symlinks) and sort them by their full path. For each entry in the contents table, compute the hash for the concatenation of: |
There was a problem hiding this comment.
We probably need to better specify what "sort" means, and in particular, what collation ordering we want to use. Because:
$ echo -e "aa\na-a\nab\na-b" | LC_ALL=en_US.UTF-8 sort
a-a
aa
a-b
ab
$ echo -e "aa\na-a\nab\na-b" | LC_ALL=C.UTF-8 sort
a-a
a-b
aa
ab
There was a problem hiding this comment.
Like Python's sort([str, ...]), it's locale agnostic. See 378d1fd (#100)
Standardizes findings in conda/conda-build#5277.
🔎 Preview in Markdown