Re: [Gnu-arch-users] Naturally "inconsistent" naming convention

[Thomas Lord]

    > From: Stefan Monnier <address@hidden>

    > Arch is based on the idea that you have to have a consistent naming
    > convention, which you should describe in =tagging-method.

That's roughly correct.

    > But what about those cases where some classes of files are source while
    > others are auto-generated.  A typical example being C files generated by
    > lex&yacc.

What about them?  Some projects want to count those generated files as
source, others don't --- arch provides enough flexability to achieve
either aim.

Sometimes people gripe that they *can't possibly* get their build
system to not stash those files in source trees.  Strangely,
confronted with that beginner's-mistake difficulty in their build
system, they choose to complain here.

    > Are we supposed to tell lex&yacc to generate file names like
    > +lexer.c (which won't work in languages like OCaml or Java where
    > where file names have to be the same as module names)?  Or are
    > we supposed to list of those exception in arch-inventory?


A less strident/flippant answer is:

The inventory system is very valuable, both within arch and without.
(Consider, for example, how `inventory' can be usefully substituted
for `find' in various pipelines.)

I would not and do not claim that the current inventory system is the
last word in inventory systems.  I would speculate that the current
inventory system can be improved.  The question is: how exactly?  I
would claim that that is a deep question and requires careful thought.

Thanks,
-t