Re: [Gnu-arch-users] automatic messages (for archive cycling etc)

[Miles Bader]

On Mon, Feb 02, 2004 at 08:17:03AM -0800, Tom Lord wrote:
> In general, there's some things that all versions should more or less
> agree on -- such =tagging-method.   But other things, such as
> ChangeLogs, where I want per-version (and thus am likely to use things
> like "ChangeLog.d" directories).   I thiink messages are in the
> per-version category.
> 
> Given that: my next question is "Why are messages a separate file?
> Why not just a log header?"
> 
> As in:
> 
>       X-version-message: This version has migrated
> 
> and things that print messages can print:
> 
> 
>       =========================================
>               This version has migrated
...
>       <log body>
>       =========================================
> 
> (i.e., print a subset of headers plus the message body, 
> formatted loudly.)

I'm not really sure what you mean.  Which log files?  Does tla have to search
through them to find messages?  Given that (as I've defined it) these
messages are _persistant_ -- they get spat out until actively canceled by
removing the file -- tla would have to do such searching constantly.

The advantage of a single file (or say multiple per-version files in a
directory) is that tla doesn't have to work very hard to deal with them.

Some other apparent points against using log files thus:

  * How would such messages be deleted once dealt with?  Would you delete the
    log file (something which has other consequences in arch, and anyway is a
    bit taboo)?  Would you add a special `cancel' header to a subsequent log
    file (adding more complicated mechanism)?

  * The example you show above suggest a rather fixed format; I rather like
    the idea of something the user can completely define, and thus use for
    occasions that might not be envisioned now (advertising space? :-)

  * I think a single-file/simple-directory-of-files would also be easier for
    users to understand: the way they would work is really, really, obvious.
    A special log header, though not rocket-science, looks like it would
    entail more rules and mechanism to explain/document.

`Why not just a log header?' seems to imply that using `a log-header' should
be simpler, but I'm not sure I see it...

-Miles
-- 
`To alcohol!  The cause of, and solution to,
 all of life's problems' --Homer J. Simpson