Case sensitivity

[Alexander Belchenko]

Talden пишет:
> Here I am being a top-posting sheep...
> 
> It's important to understand (and I'm sure Alexander understands this
> point) that this is not so much a bug, as a design decision.

I'm sorry? Why you think that I think this is a design decision?

No, this is THE bug, based on wrong assumptions of main developers.
I don't think it was "design decision", I believe it was just "works for 
me" decision. I understand why this problem exists, but I don't think it 
is intended.

>  It is
> also not an unusual limitation of VCS software either (CVS,
> Subversion, ...) - and AFAIK a universal limitation for any VCS
> intended for platform interoperability.

Svn has support for this problem, although their solution leads to "not 
preserve case at all" behavior (you can't change the filename case on 
windows with svn at all).

>  It might be worth looking to
> see if a bug has been filed (and hopefully noted as an enhancement) to
> see what other dialogue exists on the topic.

Yes, the bug filed and IIRC even not one. Look under win32 tag.

> Consider that enhancing Bazaar for this is non-trivial.

Yes, because as you called it "design decision" and it's ingrained in 
deepest basic algorithms.

>  It will
> complicate the interpretation of historical revisions (and I expect
> commit meta-data would be needed to differentiate new revisions from
> old) and face issues reverting correctly to an earlier tree.
> Additionally, this enhancement would introduce a case-related
> evil-twin problem for case-sensitive file-systems attempting to work
> with repositories that expect case-insensitivity - at the least this
> would complicate the UI.

Scary. Very scary.

Actually bzr has very good support for renames, and if you treat change 
of case as rename (even implicit) and you commit this rename explicitly, 
then there is no history hell as you describe it.

So any interaction with windows file system should be done only on a 
working tree level.

I should note that I saw vfat on linux has even weirdest bugs re case 
sensitivity.

> Personally, though I'm a Windows encumbered developer (Heavy
> Encumbered, half-speed), I think it would be better for these tools to
> stick with case-sensitive file-system as it dramatically simplifies
> implementation and only rarely do applications on case-insensitive
> file-systems change the case of files without user intervention - far
> better that those few applications are compelled to be fixed.
> 
> --
> Talden.

Alexander

PS. vila, I'm trying, I'm trying!