Complexity of mutability

[Daira Hopwood]

On 08/08/15 00:07, Lukas Pirl wrote:
> On 08/06/2015 10:56 PM, Adam Hunt wrote:
>> Something that I've been wondering since then is the complexity
>> introduced by mutable files. In a few use cases that I've been thinking
>> about mutability is unnecessary and potentially even a liability. How
>> much complexity is introduced into Tahoe's design to allow for mutabile
>> files? If mutability was eliminated from an implementation of a system
>> based on Tahoe's design would the system become appreciably less complex?
> 
> That raises a question in my head: when mutable files are mutated, are
> the existing shares updated or are new shares created?

The existing shares are updated.

> And what happens to the servers holding shares that cannot be
> [successfully] updated?

Those shares are not updated. This is why updates to mutable files
can potentially cause data loss.

> I ask because the garbage collection in Tahoe is a point where I see a
> lot of room for improvement.

Yes (I'll discuss this in another reply).

-- 
Daira Hopwood ⚥

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 490 bytes
Desc: OpenPGP digital signature
URL: <http://tahoe-lafs.org/pipermail/tahoe-dev/attachments/20150811/1b2341b8/attachment.pgp>