Re: [Monotone-devel] Re: Future of monotone

[Markus Schiltknecht]

Hi,

Thomas Moschny wrote:
> Because currently, the 'date' field is only loosely coupled to 
> the 'author', 'branch' and 'changelog' certs. If there are multiple commits 
> (e.g. in case of a clean merge), you can only guess by looking at the signers 
> what certs belong together. 

Hm.. yeah, I was just thinking about that date in general. In everyday 
live, certs normally have a date. A signing date. So, if we want to 
resemble the commonly understood

> > > Maybe one also wants to have a message' field, but I'm not sure what the
> > > use case would be, and we surely don't want to continue that ad infinitum
> > > (msg for the msg for the msg...).
> > I'd vote against such a message field, exactly for the "ad infinitum"
> > reason.
>
> It might be worth adding a short note to a tag, for example, but I am not 
> strongly after this message field.

Okay, this applies to 'branch' and 'tag'. Possibly also 'test-result', 
maybe even for 'suspend', but certainly not for 'commit-message' - which 
is by far the most frequent one...

Isn't it cleaner adding a 'comment' cert where needed. (And make sure 
that ends up close to the other cert... locality.)

> > > This would leave us with currently four pre-defined cert types: 'commit'
> > > resp. 'commit-message', 'branch', 'tag' and 'suspend'.
> > Hm.. having the date information merged into these certs would lead to
> > duplicates for normal commits.
>
> That could be avoided by additionally combining 'changelog' and 'branch' certs 
> into one cert.

Uh... right.

So in your world, there still are:

  - 'commit' (branchname => value,  changelog => comment)
  - 'tag'    (possibly with a comment)
  - 'test-result'  (possibly with a comment)

Right? Hm... so every cert would have a value, a date, a signer and 
possibly a comment.

> That's imho a different topic, and I'm not sure I already understood what you 
> mean by 'locality' here.

Well, what's your reason to want to merge these certs and their values 
together? I thought the primary reason was locality (i.e. having those 
values together, which belong together).

Regards

Markus