Re: [Gnu-arch-users] Re: tla1.2 on cygwin

[Dustin Sallings]

On Mar 10, 2004, at 1:21, Tom Lord wrote:

> >         Turns out there are quite a few UNIX and Windows users out there with
> > case insensitive filesystems
>
> Encourage them to upgrade.

	I never took you as one of those, ``you wouldn't have this problem if 
you ran Linux'' developers I kept running into a few years ago when my 
primary platforms were Solaris, IRIX, and NetBSD.

> > and this particular design decision seems to do very little
> > other than to break archives that are ever checked out and then
> > recommitted by people who use [broken file systems].
>
> >         Does the rearrangement of the structure in this part of the design
> > have any positive effects?
>
> Yes, and they've been explained.

	Can you point me to an explanation?  All I've seen is some belief that 
it makes manually digging through the patch logs for a particular 
version easier if everyone keeps category and branch names consistent.  
I can't imagine how that could be the case 
({arch}/*/*/*--1.0/*/patch-log vs. {arch}/*/*/*/*--1.0/patch-log).

> > It was brought up in this thread by a user
> > who finds it counterintuitive.  I don't believe this user is alone, 
> > but
> > I wouldn't be complaining if it hasn't caused me to actually break
> > archives when working on cross-platform, multi-user, distributed
> > applications.
>
> What's that you say?  The sky is falling?   Interesting.

	I have an actual broken archive created by a broken tool that made bad 
assumptions due to a design decision to invert the archive/category 
relationship in one part of the system and this is the type of response 
I get?

	I'm an arch evangelist.  I believe strongly in this project and I 
encourage my wife to dump money into your paypal account as the value 
that arch has brought me has greatly exceeded my expectations.  This 
type of response makes it really difficult to convince people to trust 
their work to arch.

	Yeah, the sky is falling.  My first interactive attempt working with 
other people in the way arch was designed to be used has left me with a 
completely broken archive and your response is that Microsoft and Apple 
should change fundamental properties of their operating systems to 
better accommodate the way you do things.

> Your focused on a narrow issue that will _not_ solve the interop
> problems that can arise because of case insensitive file systems.
> So, you are asking for a non-solution to a bogus problem.

	I'm focused on an issue where the tool itself breaks regardless of 
coding standards used within a project when someone on some local 
temporary archive decides to use a category name with a different case 
and get his patch submitted back in.

	If someone submits a patch to a project where they try to add an X.h 
in the same directory as my x.h, that's easy to detect and refuse (and 
I would just because it's confusing).  It's quite a bit different (and 
rather hard) to go back and consider the lineage of a patch to figure 
out if any developer ever used a branch name that conflicted with a 
branch name that some other developer used.

--
SPY                      My girlfriend asked me which one I like better.
pub  1024/3CAE01D5 1994/11/03 Dustin Sallings <address@hidden>
|    Key fingerprint =  87 02 57 08 02 D0 DA D6  C8 0F 3E 65 51 98 D8 BE
L_______________________ I hope the answer won't upset her. ____________