Mercurial log between two tags

Question

I want to use hg log to generate a brief changelog covering the changes of the latest release. Releases are tagged with a "v" prefix e.g. "v0.9.1" or "v1.0". Is it possible with revsets to select a range between the two last tags that start with "v", not including the older one of the two tags?

As an example if I have:

• changeset 0
• changeset 1
  tag: v0.1
• changeset 2
• changeset 3
  tag: bar
• changeset 4
  tag: v0.9
• changeset 5
  tag: foo
• changeset 6
• changeset 7
  tag: v1.0
• changeset 8

I'd like to select changesets 5 to 7, if possible without even specifying v0.9 and v1.0 only using the "v" prefix.

Answer 1

IF you really want the answer to the question, "what changed between two revisions" you have to do something a little verbose:

hg log -r 'ancestors(relB) - ancestors(relA)'

This is different than relA::relB in the case where there was branched development.

Consider the following history graph:

$ hg glog
@  changeset:   7:cb998461fc2f
|  tag:         tip
|  user:        Danny Sadinoff <danny>
|  date:        Fri Nov 21 09:28:46 2014 +0000
|  summary:     Added tag relB for changeset b37a8f32de1d
|
o    changeset:   6:b37a8f32de1d
|\   tag:         relB
| |  parent:      5:187092bb7e9b
| |  parent:      3:2d5af4e62ad5
| |  user:        Danny Sadinoff <danny>
| |  date:        Fri Nov 21 09:27:47 2014 +0000
| |  summary:     merged
| |
| o  changeset:   5:187092bb7e9b
| |  user:        Danny Sadinoff <danny>
| |  date:        Fri Nov 21 09:27:21 2014 +0000
| |  summary:     more branch work
| |
| o  changeset:   4:dbfe8f003e51
| |  parent:      1:68f933b431ec
| |  user:        Danny Sadinoff <danny>
| |  date:        Fri Nov 21 09:26:51 2014 +0000
| |  summary:     branched dev
| |
o |  changeset:   3:2d5af4e62ad5
| |  user:        Danny Sadinoff <danny>
| |  date:        Fri Nov 21 09:26:17 2014 +0000
| |  summary:     Added tag relA for changeset 0ad160ab1455
| |
o |  changeset:   2:0ad160ab1455
|/   tag:         relA
|    user:        Danny Sadinoff <danny>
|    date:        Fri Nov 21 09:25:13 2014 +0000
|    summary:     mainline dev
|
o  changeset:   1:68f933b431ec
|  user:        Danny Sadinoff <danny>
|  date:        Fri Nov 21 09:24:18 2014 +0000
|  summary:     added a line
|
o  changeset:   0:4738f4f68f1e
   user:        Danny Sadinoff <danny>
   date:        Fri Nov 21 09:23:54 2014 +0000

observe the difference between @Kevin's result and the ancestors() approach:

$ hg log -r 'ancestor(relA, relB)::relB - ancestor(relA, relB)'
changeset:   3:2d5af4e62ad5
user:        Danny Sadinoff <danny>
date:        Fri Nov 21 09:26:17 2014 +0000
summary:     Added tag relA for changeset 0ad160ab1455

changeset:   6:b37a8f32de1d
tag:         relB
parent:      5:187092bb7e9b
parent:      3:2d5af4e62ad5
user:        Danny Sadinoff <danny>
date:        Fri Nov 21 09:27:47 2014 +0000
summary:     merged

versus the ancestors approach, which tells the whole story.

$ hg log -r 'ancestors(relB)  - ancestors(relA)'
changeset:   3:2d5af4e62ad5
user:        Danny Sadinoff <danny>
date:        Fri Nov 21 09:26:17 2014 +0000
summary:     Added tag relA for changeset 0ad160ab1455

changeset:   4:dbfe8f003e51
parent:      1:68f933b431ec
user:        Danny Sadinoff <danny>
date:        Fri Nov 21 09:26:51 2014 +0000
summary:     branched dev

changeset:   5:187092bb7e9b
user:        Danny Sadinoff <danny>
date:        Fri Nov 21 09:27:21 2014 +0000
summary:     more branch work

changeset:   6:b37a8f32de1d
tag:         relB
parent:      5:187092bb7e9b
parent:      3:2d5af4e62ad5
user:        Danny Sadinoff <danny>
date:        Fri Nov 21 09:27:47 2014 +0000
summary:     merged

the actual "greatest common revision" between two revisions seems to be:

 ancestor(ancestors(relB)  - ancestors(relA))

so a pretty glog of the changes becomes:

$ hg glog -r 'ancestor(ancestors(relB)  - ancestors(relA))::  relB'
o    changeset:   6:b37a8f32de1d
|\   tag:         relB
| |  parent:      5:187092bb7e9b
| |  parent:      3:2d5af4e62ad5
| |  user:        Danny Sadinoff <[email protected]>
| |  date:        Fri Nov 21 09:27:47 2014 +0000
| |  summary:     merged
| |
| o  changeset:   5:187092bb7e9b
| |  user:        Danny Sadinoff <[email protected]>
| |  date:        Fri Nov 21 09:27:21 2014 +0000
| |  summary:     more branch work
| |
| o  changeset:   4:dbfe8f003e51
| |  parent:      1:68f933b431ec
| |  user:        Danny Sadinoff <[email protected]>
| |  date:        Fri Nov 21 09:26:51 2014 +0000
| |  summary:     branched dev
| |
o |  changeset:   3:2d5af4e62ad5
| |  user:        Danny Sadinoff <[email protected]>
| |  date:        Fri Nov 21 09:26:17 2014 +0000
| |  summary:     Added tag relA for changeset 0ad160ab1455
| |
o |  changeset:   2:0ad160ab1455
|/   tag:         relA
|    user:        Danny Sadinoff <[email protected]>
|    date:        Fri Nov 21 09:25:13 2014 +0000
|    summary:     mainline dev
|
o  changeset:   1:68f933b431ec
|  user:        Danny Sadinoff <[email protected]>
|  date:        Fri Nov 21 09:24:18 2014 +0000
|  summary:     added a line
|

Answer 2

If you know the names of the tags off the top of your head:

hg log -r 'v0.9::v1.0 - v0.9'

If you don't, you can find out:

hg log -r 'last(tag("re:^v"), 2)'

The first log command only makes sense if v0.9 is an ancestor of v1.0; if this is not the case, you can use a single colon instead of ::, but the result will probably be total nonsense. In that scenario, you may want something like this instead:

hg log -r 'ancestor(v0.9, v1.0)::v1.0 - ancestor(v0.9, v1.0)'

That will produce a log of all changes starting from the most recent common ancestor of v0.9 and v1.0. If v0.9 is an ancestor of v1.0, this degenerates to the same behavior as the first log command.