[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Gnu-arch-users] Re: tagline robustness
From: |
Tom Lord |
Subject: |
[Gnu-arch-users] Re: tagline robustness |
Date: |
Wed, 27 Aug 2003 08:28:20 -0700 (PDT) |
> From: Miles Bader <address@hidden>
> Tom Lord <address@hidden> writes:
> > The original architectural conception here is of inventory being
> > reasonably (and usefully) separable from the rest; mkpatch/dopatch
> > being similarly factorable-outable (but layered on inventory, of
> > course); and then patch-logs, the namespace, and the core archive
> > protocol being a layer on that.
> I've noticed this, and wondered whether it had something to do with the
> original implementation being via shell-scripts -- certainly this
> property of arch has made my own little shell scripts often very easy
> to write.
It's hard to go back and dissect my thinking this much later but:
1) The layering/software-tools-approach I described came first.
For example, I briefly considered implementing it in something
SCSH-like first -- but with the same architecture.
2) The shell-based implementation helped to keep the architecture
"honest" -- it's hard to _not_ stick to a software-tools approach
if you want to keep your shell code clean. Conversely, it's
harder now, with tla, to keep the architecture clean.
3) A purist might argue that the point at which things started going
to hell was when `inventory' was changed from a shell script that
invoked `find' (with several dozen arguments) to a stand-alone
program that did it's own traversal. The current plans for
generalizing =tagging-method really drive that point home: it's
already the case and soon much more so that you _can't_ compute
a tree inventory other than by using this very specific, highly
configurable tool.
I'm not sure that could have been avoided. _I_ would have thought
"Ok, sure, a very limited `inventory', such as could be implemented
with `find' and `sed', really suggests using an enhanced discipline
about how you layout your source trees -- but that's a good thing."
As you can see from list archives, though, it's too much of a
religious issue with too many people.
[other replies separately]
-t
- Re: [Gnu-arch-users] Re: tagline robustness, (continued)
- Re: [Gnu-arch-users] Re: tagline robustness, Miles Bader, 2003/08/24
- Re: [Gnu-arch-users] Re: tagline robustness, Jan Hudec, 2003/08/25
- [Gnu-arch-users] Re: tagline robustness, Miles Bader, 2003/08/25
- Re: [Gnu-arch-users] Re: tagline robustness, Jonathan Walther, 2003/08/26
- Re: [Gnu-arch-users] Re: tagline robustness, Miles Bader, 2003/08/26
- Re: [Gnu-arch-users] Re: tagline robustness, Adam Sampson, 2003/08/26
- Re: [Gnu-arch-users] Re: tagline robustness, Tom Lord, 2003/08/26
- [Gnu-arch-users] Re: tagline robustness, Miles Bader, 2003/08/26
- Re: [Gnu-arch-users] Re: tagline robustness, Tom Lord, 2003/08/26
- [Gnu-arch-users] Re: tagline robustness, Miles Bader, 2003/08/26
- [Gnu-arch-users] Re: tagline robustness,
Tom Lord <=
- Re: [Gnu-arch-users] Re: tagline robustness, Jason McCarty, 2003/08/27
- [Gnu-arch-users] Re: tagline robustness, Miles Bader, 2003/08/27
- [Gnu-arch-users] Re: tagline robustness, Maksim Lin, 2003/08/28
- Re: [Gnu-arch-users] Re: tagline robustness, Robert Anderson, 2003/08/28
- Re: [Gnu-arch-users] Re: tagline robustness, Zack Brown, 2003/08/28
- Re: [Gnu-arch-users] Re: tagline robustness, Jason McCarty, 2003/08/28
- [Gnu-arch-users] file and directory restrictions, Zack Brown, 2003/08/28
- [Gnu-arch-users] Re: tagline robustness, Miles Bader, 2003/08/28
- [Gnu-arch-users] Re: tagline robustness, Jason McCarty, 2003/08/28
- [Gnu-arch-users] towards standards specifications, Tom Lord, 2003/08/27