[OpenAFS] Re: OpenAFS Project List

Derrick J Brashear shadow@dementia.org
Wed, 14 Mar 2001 23:10:32 -0500 (EST)


On 14 Mar 2001, Johan Danielsson wrote:

> "Laura Stentz" <stentz@us.ibm.com> writes:
> 
> > |------------------+--------------+---------------+----------------------|
> > | @sys expansion   |              | High          | Derrick Brashear will|
> > |                  |              |               | write up the proposal|
> > |                  |              |               | of "best practices   |
> > |                  |              |               | for using sysname".  |
> > |                  |              |               | Suggested naming     |
> > |                  |              |               | convention is        |
> > |                  |              |               | i386_openbsd_28_28.  |
> > |                  |              |               | (Craig Everhart to   |
> > |                  |              |               | have IBM/AFS         |
> > |                  |              |               | development to       |
> > |                  |              |               | comment on the       |
> > |                  |              |               | proposal to determine|
> > |                  |              |               | if IBM will follow   |
> > |                  |              |               | the openafs          |
> > |                  |              |               | convention.)         |
> > |------------------+--------------+---------------+----------------------|
> 
> Since it's `good practice' to include the sysname in per architecture
> volumes, this pretty much requires that volume names can be longer
> than 32 characters.

The argument was made that short names be used for volumes, while full
names would be used for sysname values; Additionally, existing sysname
families would be grandfathered.

> And I'm also not so sure you can come up with a universal naming
> scheme, for instance on the various BSD's the machine type makes a
> pretty good first component, but it might not be the same as used on
> other systems.

-D