[tahoe-dev] Weekly Dev Chat notes from 2013-03-07

[Zooko O'Whielacronx]

On Mon, Mar 11, 2013 at 12:26 PM, Greg Troxel <gdt at ir.bbn.com> wrote:
> To me, this seems like needless churn.  I don't think the current scheme is broken.

Noted! Thanks for the feedback.

The advantage of $YEAR.$WHATEVER versioning is that then people who
don't have any memory of Tahoe-LAFS history, nor the time to look it
up, can get an idea of what "vintage" of software we're talking about.
Seems to work pretty well for Twisted and Ubuntu.

> It would be cool for someone to set up to run inter-node regression tests of various versions.

Yes, that would! I value backward-compatibility, because I recognize
that people are often more conservative about storage than about other
kinds of software.

There used to be a thing called Memory Lane:

https://tahoe-lafs.org/trac/tahoe-lafs/wiki/TestGrid?version=54

But it needs a maintainer:

https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1002# do some
renovations on Memory Lane

That would be a pretty cool thing to have on the live Test Grid, so if
you had a question like "What happens if you upload a new
XSalsa20-encrypted file using Tahoe-LAFS v14.1 to a grid with one
Tahoe-LAFS storage server of each version going back a few years:
1.10.0, 1.9.2, 1.8.3, 1.7.1, 1.6.1, 1.5, 1.4.1, 1.3, 1.2, 1.1, and
1.0, what happens?", then you can just try it and see! (Hopefully the
answer is It Just Works because the storage servers don't know what
kind of encryption you are using on the client side. ☺)

Regards,

Zooko