[PATCH] hgweb: display difference for a changeset against any parents (issue2810)

[Matt Mackall]

On Mon, 2012-11-19 at 18:49 +0000, Weiwen Liu wrote:
> Yes, y is always the same as z.
> 
> The challenge is that I can't find x when trying to display diff/x:y using
> 
> {baseline%changesetbaseline}
> 
> changesetbaseline = '<a
> href="{url}rev/{originalnode|short}:{node|short}{sessionvars%urlparameter}"
> >{node|short} </a> '
> 
> Baseline is to walk all parents and it replaces original node with the
> current parent node value.
> 
> 
> Is there a way to retrieve original node without passing it through
> template variables?

No. The point of my mail is that I agree that you need to pass more to
the templates. But ONLY one more datum: x.

> Thanks,
> Weiwen
> 
> 
> On 11/17/12 12:25 PM, "Matt Mackall" <mpm at selenic.com> wrote:
> 
> >On Fri, 2012-11-16 at 19:28 +0000, Weiwen Liu wrote:
> >> Thanks for feedback.
> >> 
> >> On URL scheme, I like your suggestions.  diff/<node a>:<node b>
> >> extends diff command instead of introducing a new one.  And it follows
> >> more closely pattern in the NewWebInterface (NewWebInterface considers
> >> anything beyond the first node as file path). I will try this one out.
> >> 
> >> Conditional template is interesting.
> >> Two new template variables are used: original node and baseline.
> >> With expansion of diff command, it makes it easier to know which
> >> baseline is used by displaying baseline info in the page.
> >> If a diff is against a node's only parent, we can skip the link for
> >> changing baseline.  If a diff is against a node's ancestor, or sibling
> >> from a different branch, it is convenient to have a link to see
> >> difference against direct parent.
> >> I prefer to keep these new variables for the purpose of consistency and
> >>simplicity in code.  Let me know if this makes sense.
> >
> >The template needs the following information:
> >
> >a) which two revs (x, y) are involved in the diff
> >b) the bits relevant to the "header" revision z
> >
> >I'm pretty sure it makes sense for y to always be the same as z, so I
> >think the only additional information needed is x (usually the same as
> >p1(y)).
> >
> >Note that the "list of choices of diff base" are already inherent in b
> >as p1 and p2.
> >
> >-- 
> >Mathematics is the supreme nostalgia of our time.
> >
> >


-- 
Mathematics is the supreme nostalgia of our time.