On Wed, May 11, 2005 at 02:06:01AM -0500, Brad Appleton wrote:
> I think of a mnemonic name for a revision as something different from
> the sort of first-class label some are speaking of (one that possibly
> requires versioning of the configuration of the label, in order to be
> "TimeSafe").
The first-class label referred to seem to be more-or-less a
reincarnation of the CVS tag. As I've said before, I don't see any
particular advantage to have such things, since the current cheap-copy
mechanism in Subversion already suffices to deal with this. However,
this is something different from having labels to refer to revision
numbers.
> The mnenomic is simply an alias. I could possibly even define more than
> one for the same global revnum (provided the name isnt already in use).
> The purpose isnt to define the content of the configuration
> corresponding to the revnum, the purpose is to be able to use a
> meaningful, memorable name for a state of the codebase that is
> automatically stored+tracked by the system anytime I checkin.
Yes, this is what I had in mind as well. To be precise, I consider the
mnemonic-for-revision feature to be a function (in the mathematical
sense) from a set of human-readable strings to revision numbers. This
mapping need not be, and should not be required to be, one-to-one. It
is not bound to any particular URL within the repository. It is simply
a label referring to the state of the repository at a particular time,
which I happen to consider to be special enough an occasion to deserve
a mnemonic name rather than a plain revision number. I'd like to be
able to refer to this special point in the repository's history by
using the name rather than the number.
> With "tags" or "labels" as most tools implement them, the purpose IS to
> define the content of a configuration of the system that is otherwise
> NOT auto-captured, *and* to associate it with a meaningful name. I think
> there is a use for that as a separate thing in SVN, and that "copies"
> with some read-only flag set serve that purpose just fine.
Agreed.
> A traditional "label" manually creates a named configuration.
> A revision alias simply associates a name with an autogenerated
> id/address for an automatically created+stored configuration.
>
> I think they are different things, just like the list of symbol names in
> a symbol-table is different from the actual data/memory values that are
> referenced by the addresses that the "symbols" map to.
[...]
Agreed 100%.
I hope people won't wrongly conflate the two, and reject the mnemonic
name for revision idea just because the *other* idea is flawed!
T
--
Klein bottle for rent ... inquire within. -- Stephen Mulraney
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Wed May 11 20:44:18 2005