Re: [Gnu-arch-users] GNU Arch review - am I accurate?

[David A. Wheeler]

> On Wed, 2004-03-03 at 01:07, David A. Wheeler wrote:
> > There are some things I didn't see:
> > * Is anyone currently working on automated caching?
> 
> Of course -- Arch has automated caching build in. See "revision
> libraries", archive mirrors and cacherevs. There've been some folks
> proposing caching other files as well, but such proposals typically add
> nothing revlibs don't do already.
> 
> > * Is it even slightly plausible to change the default
> >   filename/tagname conventions so arch will
> >   work more easily with common tools (e.g., vi/vim, more, csh,
> >   bash, Windows (it doesn't handle long names well))?
> >   Conventions are so arbitrary, yet the ones arch uses
> >   seem designed to cause unnecessary problems.
> 
> Go read the archives. This has been rehashed time and time again

Any suggestions on keywords to search for?
Searches for "naming convention" and such don't see to pull
that sort of thing.


> -- but
> folks who actually try to use Arch for a while (myself included) tend to
> appreciate the conventions as they stand.

My problem isn't with conventions per se, it's that the
default conventions interfere with some common tools.
So, are there conventions that could be used instead?

Several have commented that they want the marker character
first, so that they sort specially.  Frankly, I don't know
if that's such a great idea - if you want them at the beginning,
you could use capitals letters or whatever.  But okay, if
that's important let's look for a different character that
can be used at the beginning of a filename.  Most characters
have some sort of problem, but here are a few candidates:
^ this was the old Unix pipe symbol before |, but almost no
  system accepts that as a pipe anymore.  Works on Windows.
  On some shells (like C shell), you'd have to prefix with ./
  to execute it, but since people's PATH shouldn't include "."
  that would be true anyway.
% This is an argument prefix on Windows (e.g., %1), so it
  may be problematic on Windows systems.
@ This is used for email and in URLs to identify usernames.
  Still, it's not a bad idea.
_ It looks like "-", which gives the risk of typing the wrong one,
  but it seems to work very well.

So at least ^ seems to be easy to use, and address@hidden are plausible as well.
Windows XP doesn't barf on any of those characters as a leading character.


> Making a practice of changing longstanding conventions easily is also
> not a friendly thing to do with software in production usage.

On THAT we agree!  But that can be handled by just
making sure that the regex'es for tagging include the
"original" and "new" convention.  If it affects
the internals, modify the code slightly so
it can handle the old and new ways too
("look for new way, then look for old way, THEN declare it's not there").

For naming conventions, I think that

--- David A. Wheeler <address@hidden>