mount on case insensitive filesystems

[jbenet]

Some filesystems (like osx's default hfs) are case insensitive... this breaks our b58 string keys when mounting with fuse. Possible solutions

• use different encoding (argh)
• use a different mount point to map between encodings /ipfs/b32/...

[whyrusleeping]

I... I dont... why? A unix based OS with case insensitive paths? I hurt...

[chriscool]

Yeah, there is also the way unicode is handled in hfs. See: http://radsoft.net/rants/20080405,00.shtml
I don't remember what was eventually done in Git about this.

[whyrusleeping]

@chriscool if you could figure out how git handles it, that would be super cool

[chriscool]

@whyrusleeping ok, first here are some related files:

https://github.com/git/git/blob/master/t/t0050-filesystem.sh
https://github.com/git/git/blob/master/t/t3910-mac-os-precompose.sh
https://github.com/git/git/blob/master/compat/precompose_utf8.h
https://github.com/git/git/blob/master/compat/precompose_utf8.c

I will dig deeper later.

[chriscool]

This commit is also interesting:

git/git@76759c7

[chriscool]

It looks like go has good support for unicode normalization:
http://blog.golang.org/normalization

[jbenet]

Thanks @chriscool all of these are very interesting finds!

[chriscool]

Thanks @jbenet.

By the way, in https://github.com/git/git/blob/master/builtin/init-db.c around line 288 there is:

                /* Check if the filesystem is case-insensitive */
                path[len] = 0;
                strcpy(path + len, "CoNfIg");
                if (!access(path, F_OK))
                        git_config_set("core.ignorecase", "true");
                probe_utf8_pathname_composition(path, len);

and in https://github.com/git/git/blob/master/git-compat-util.h there is:

/* used on Mac OS X */
#ifdef PRECOMPOSE_UNICODE
#include "compat/precompose_utf8.h"
#else
#define precompose_str(in,i_nfd2nfc)
#define precompose_argv(c,v)
#define probe_utf8_pathname_composition(a,b)
#endif

[chriscool]

About the core.ignorecase config variable it is read in config.c into the ignore_case global variable and then in dir.c there is:

/* helper string functions with support for the ignore_case flag */
int strcmp_icase(const char *a, const char *b)
{
        return ignore_case ? strcasecmp(a, b) : strcmp(a, b);
}

int strncmp_icase(const char *a, const char *b, size_t count)
{
        return ignore_case ? strncasecmp(a, b, count) : strncmp(a, b, count);
}

int fnmatch_icase(const char *pattern, const char *string, int flags)
{
        return wildmatch(pattern, string,
                         flags | (ignore_case ? WM_CASEFOLD : 0),
                         NULL);
}

[mitar]

hfs is case insensitive, but case preserving. You can still store files with different case, you just cannot have two files which have name different only in case.

[Stebalien]

CIDs fix this as we can use any base we want. Also, we're switching to base32 by default so this will become even less of an issue.