You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Heikki Orsila <sh...@jolt.modeemi.cs.tut.fi> on 2007/09/05 14:40:03 UTC

git log vs. svn log

Hello SVN developers.. I'm greatly missing two features from svn log 
command that are featured in git log:

1. git log --stat shows commit diffstats
2. git log -p shows commit patches

Two examples should clarify these options:

$ git log --stat
commit 795688a4905257516886cac6f3add6ba467dfe0b
Author: Heikki Orsila <sh...@thingol.localdomain>
Date:   Wed Aug 22 23:42:24 2007 +0300

    Add documentation of -i and -o [DOCUMENTATION]

 pmr.1 |   16 ++++++++++++++--
 pmr.c |   15 +++++++++++++--
 2 files changed, 27 insertions(+), 4 deletions(-)

$ git log -p

commit 795688a4905257516886cac6f3add6ba467dfe0b
Author: Heikki Orsila <sh...@thingol.localdomain>
Date:   Wed Aug 22 23:42:24 2007 +0300

    Add documentation of -i and -o [DOCUMENTATION]

diff --git a/pmr.1 b/pmr.1
index f42235a..d3fd6d2 100644
--- a/pmr.1
+++ b/pmr.1
@@ -1,11 +1,11 @@
-.TH pmr 1 "2006-11-03" Linux "user commands"
+.TH pmr 1 "2007-08-22" Linux "user commands"
...

These two options really made version control logs useful for me, so I 
really hope to see these features in SVN in the future (I can not 
use git at work). It's annoying to live without them ;) Note, I'm not 
switching to version control GUIs, I want to use command line tools.

-- 
Heikki Orsila			Barbie's law:
heikki.orsila@iki.fi		"Math is hard, let's go shopping!"
http://www.iki.fi/shd

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: git log vs. svn log

Posted by Daniel Rall <dl...@collab.net>.
On Wed, 05 Sep 2007, Heikki Orsila wrote:
...
> $ git log --stat
> commit 795688a4905257516886cac6f3add6ba467dfe0b
> Author: Heikki Orsila <sh...@thingol.localdomain>
> Date:   Wed Aug 22 23:42:24 2007 +0300
> 
>     Add documentation of -i and -o [DOCUMENTATION]
> 
>  pmr.1 |   16 ++++++++++++++--
>  pmr.c |   15 +++++++++++++--
>  2 files changed, 27 insertions(+), 4 deletions(-)
...

I don't personally see a whole lot of value in these diffstats past
what can be gotten from 'svn log --verbose'.

- Dan

p.s. Heikki, thanks for the feedback.

Re: git log vs. svn log

Posted by Malcolm Rowe <ma...@farside.org.uk>.
On Wed, Sep 05, 2007 at 03:30:25PM -0400, C. Michael Pilato wrote:
> Ideally, the Subversion repository would house this type of contextual diff
> information so that, for example, we could transmitting full binary diff

I assume there's a missing "avoid" there?  ("could avoid transmitting").

> streams across the wire when running the links the 'svn blame'.  But it
> doesn't.  The primary storage format in Subversion's repository is a binary
> diff that is line-ignorant, and therein lies the primary reason why
> Subversion, unlike CVS, ain't so snappy about reporting line-change info.
> 

Regards,
Malcolm

Re: git log vs. svn log

Posted by "C. Michael Pilato" <cm...@collab.net>.
Mattias Engdegård wrote:
> shd@jolt.modeemi.cs.tut.fi (Heikki Orsila) writes:
> 
>> pmr.1 |   16 ++++++++++++++--
>> pmr.c |   15 +++++++++++++--
>> 2 files changed, 27 insertions(+), 4 deletions(-)
> 
> I wanted this information (added and removed lines) as well, so I wrote
> a script to calculate it as a post-commit hook and save the information
> in a revprop.

FWIW, ViewVC provides a similar script (svndbadmin), but instead drops the
calculated results into a MySQL Bonsai-like database.

Ideally, the Subversion repository would house this type of contextual diff
information so that, for example, we could transmitting full binary diff
streams across the wire when running the links the 'svn blame'.  But it
doesn't.  The primary storage format in Subversion's repository is a binary
diff that is line-ignorant, and therein lies the primary reason why
Subversion, unlike CVS, ain't so snappy about reporting line-change info.

-- 
C. Michael Pilato <cm...@collab.net>
CollabNet   <>   www.collab.net   <>   Distributed Development On Demand


Re: git log vs. svn log

Posted by Mattias Engdegård <ma...@virtutech.se>.
shd@jolt.modeemi.cs.tut.fi (Heikki Orsila) writes:

> pmr.1 |   16 ++++++++++++++--
> pmr.c |   15 +++++++++++++--
> 2 files changed, 27 insertions(+), 4 deletions(-)

I wanted this information (added and removed lines) as well, so I wrote
a script to calculate it as a post-commit hook and save the information
in a revprop. This works well, but:

* Counting the +/- lines of the svn diff output takes care of most cases, but
  files in deleted directories will not show up at all so they have to be
  counted explicitly.

* In svn diff, copies will count as additions, even if nothing was changed. 
  It is also usually not very interesting, especially for whole-tree copies.
  I treat tags and branches specially using path name heuristics.

For bigger changes this can take quite some time.
Perhaps the svn client could do a better job, but it is not obvious how.

Script available on request. I could submit it to contrib/, but I'm not sure
it is clean or general enough.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: git log vs. svn log

Posted by Daniel Rall <dl...@collab.net>.
On Wed, 05 Sep 2007, Ben Collins-Sussman wrote:

> On 9/5/07, Malcolm Rowe <ma...@farside.org.uk> wrote:
> 
> > I must admit, I'd find this useful as well.
> >
> 
> Historically, we've used tools/client-side/showchange.pl to do this.
> But the fact that the script exists... hey, it might be nice to have
> it as a built-in feature.  I think perforce has it too.

svnmerge.py's 'integrated' and 'avail' commands support this type of
output, too.  We've discussed providing similar options for the
upcoming Merge Tracking audit/query APIs.

Re: git log vs. svn log

Posted by Ben Collins-Sussman <su...@red-bean.com>.
On 9/5/07, Malcolm Rowe <ma...@farside.org.uk> wrote:

> I must admit, I'd find this useful as well.
>

Historically, we've used tools/client-side/showchange.pl to do this.
But the fact that the script exists... hey, it might be nice to have
it as a built-in feature.  I think perforce has it too.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: git log vs. svn log

Posted by Malcolm Rowe <ma...@farside.org.uk>.
On Mon, Sep 10, 2007 at 04:53:29PM -0700, Eric Gillespie wrote:
> Malcolm Rowe <ma...@farside.org.uk> writes:
> > I'd only want to do that kind of filtering on a feature branch, in which
> > I wouldn't expect any 'meaningful' (in the context of reviewing the
> > branch) changes to be the result of merges (in this case, presumably
> > =66rom trunk; unless I'm missing something and there would typically be
> > other merge sources too?).
> 
> Unless I branch off my feature branch to do some other sub-task
> and then merge back onto my feature branch (happens all the
> time).  I think the way to implement this feature is by taking
> one or more ignore options, e.g.  '--ignore-merges-from /trunk'.
> 

That sounds eminently sensible.

Regards,
Malcolm

Re: git log vs. svn log

Posted by Heikki Orsila <sh...@jolt.modeemi.cs.tut.fi>.
On Thu, Sep 06, 2007 at 10:02:46AM -0700, Daniel Rall wrote:
> like '--with-diff' as the option name.

That could be a long option name, but as a short option I'd preferably 
have -p, as a common practice is easier to remember. I regularly use
3 different version control systems, probably more in the future.

-- 
Heikki Orsila			Barbie's law:
heikki.orsila@iki.fi		"Math is hard, let's go shopping!"
http://www.iki.fi/shd

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: git log vs. svn log

Posted by Eric Gillespie <ep...@pretzelnet.org>.
Malcolm Rowe <ma...@farside.org.uk> writes:

> I'd only want to do that kind of filtering on a feature branch, in which
> I wouldn't expect any 'meaningful' (in the context of reviewing the
> branch) changes to be the result of merges (in this case, presumably
> =66rom trunk; unless I'm missing something and there would typically be
> other merge sources too?).

Unless I branch off my feature branch to do some other sub-task
and then merge back onto my feature branch (happens all the
time).  I think the way to implement this feature is by taking
one or more ignore options, e.g.  '--ignore-merges-from /trunk'.

--  
Eric Gillespie <*> epg@pretzelnet.org

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: git log vs. svn log

Posted by Malcolm Rowe <ma...@farside.org.uk>.
On Thu, Sep 06, 2007 at 10:02:46AM -0700, Daniel Rall wrote:
> > (For extra credit, I'd like a way to skip changes that were the result
> > of merges - can we do that already?).
> 
> 'log -g' now indicates which changes were the result of a merge
> (thanks to Hyrum's GSoC project!).  This info could be used to perform
> the filtering you describe.  However, you might filter out meaningful
> changes in the process, so I'm somewhat leary of this suggestion...

I'd only want to do that kind of filtering on a feature branch, in which
I wouldn't expect any 'meaningful' (in the context of reviewing the
branch) changes to be the result of merges (in this case, presumably
from trunk; unless I'm missing something and there would typically be
other merge sources too?).

Regards,
Malcolm

Re: git log vs. svn log

Posted by Daniel Rall <dl...@collab.net>.
On Thu, 06 Sep 2007, Malcolm Rowe wrote:

> On Thu, Sep 06, 2007 at 04:13:39PM +0200, Charles Acknin wrote:
> > On 9/5/07, Heikki Orsila <sh...@jolt.modeemi.cs.tut.fi> wrote:
> > > Filed the issue.
> > 
> > I think 'svn diff' is a more (the most?) appropriate sub-command to
> > display the diffstat stuff, rather than 'svn log'.
> > 
> > As for "2. git log -p shows commit patches", I'm not sure whether it
> > would be most relevant in 'svn diff' or 'svn log' as it displays both
> > unidiffs and commit metadata, but I'm leaning towards 'svn diff' since
> > (a) the output is mostly made of unidiffs (b) the intent is to display
> > a patch, as git-log(1) says (and as it looks like).
> > 
> 
> Right, but not necessarily a patch that you could apply - more a patch
> for review.  There's no reason you couldn't (for example) run 'svn log
> --stop-on-copy --show-unidiff .../branches/foo' to review all work done
> on a branch so far, as individual changes.

As Heikki, I like 'svn log' as the home for this functionality, and
like '--with-diff' as the option name.


> (For extra credit, I'd like a way to skip changes that were the result
> of merges - can we do that already?).

'log -g' now indicates which changes were the result of a merge
(thanks to Hyrum's GSoC project!).  This info could be used to perform
the filtering you describe.  However, you might filter out meaningful
changes in the process, so I'm somewhat leary of this suggestion...

Re: git log vs. svn log

Posted by Malcolm Rowe <ma...@farside.org.uk>.
On Thu, Sep 06, 2007 at 04:13:39PM +0200, Charles Acknin wrote:
> On 9/5/07, Heikki Orsila <sh...@jolt.modeemi.cs.tut.fi> wrote:
> > Filed the issue.
> 
> I think 'svn diff' is a more (the most?) appropriate sub-command to
> display the diffstat stuff, rather than 'svn log'.
> 
> As for "2. git log -p shows commit patches", I'm not sure whether it
> would be most relevant in 'svn diff' or 'svn log' as it displays both
> unidiffs and commit metadata, but I'm leaning towards 'svn diff' since
> (a) the output is mostly made of unidiffs (b) the intent is to display
> a patch, as git-log(1) says (and as it looks like).
> 

Right, but not necessarily a patch that you could apply - more a patch
for review.  There's no reason you couldn't (for example) run 'svn log
--stop-on-copy --show-unidiff .../branches/foo' to review all work done
on a branch so far, as individual changes.

(For extra credit, I'd like a way to skip changes that were the result
of merges - can we do that already?).

Regards,
Malcolm

Re: git log vs. svn log

Posted by Peter Samuelson <pe...@p12n.org>.
[Heikki Orsila]
> I want to see change log message (meta information) with the patch,
> so I don't think svn diff is the correct place. svn diff is used to
> compare two states of the tree, but svn log -p displays change sets
> (patches) in serial form, so they are used for totally different
> things..

On the other hand, now that we have the -c flag, I think it's
officially OK to pretend that the commands that deal with tree state
also deal with changesets.  Even though svn is fundamentally
snapshot-based rather than changeset-based, changesets can be derived
from snapshots, so I don't see why end users should have to care all
that much which is the "native" representation.  Whether the diff
subcommand has to calculate changesets from snapshots, or construct
snapshots from changesets, should be considered an implementation
detail.
-- 
Peter Samuelson | org-tld!p12n!peter | http://p12n.org/

Re: git log vs. svn log

Posted by Heikki Orsila <sh...@jolt.modeemi.cs.tut.fi>.
On Thu, Sep 06, 2007 at 04:13:39PM +0200, Charles Acknin wrote:
> On 9/5/07, Heikki Orsila <sh...@jolt.modeemi.cs.tut.fi> wrote:
> > Filed the issue.
> 
> I think 'svn diff' is a more (the most?) appropriate sub-command to
> display the diffstat stuff, rather than 'svn log'.

No, the point was to get good overall picture of changes from the log..

> As for "2. git log -p shows commit patches", I'm not sure whether it
> would be most relevant in 'svn diff' or 'svn log' as it displays both
> unidiffs and commit metadata, but I'm leaning towards 'svn diff' since
> (a) the output is mostly made of unidiffs (b) the intent is to display
> a patch, as git-log(1) says (and as it looks like).

I want to see change log message (meta information) with the 
patch, so I don't think svn diff is the correct place. svn diff is used 
to compare two states of the tree, but svn log -p displays change sets 
(patches) in serial form, so they are used for totally different 
things..

-- 
Heikki Orsila			Barbie's law:
heikki.orsila@iki.fi		"Math is hard, let's go shopping!"
http://www.iki.fi/shd

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: git log vs. svn log

Posted by Charles Acknin <ch...@gmail.com>.
On 9/5/07, Heikki Orsila <sh...@jolt.modeemi.cs.tut.fi> wrote:
> Filed the issue.

I think 'svn diff' is a more (the most?) appropriate sub-command to
display the diffstat stuff, rather than 'svn log'.

As for "2. git log -p shows commit patches", I'm not sure whether it
would be most relevant in 'svn diff' or 'svn log' as it displays both
unidiffs and commit metadata, but I'm leaning towards 'svn diff' since
(a) the output is mostly made of unidiffs (b) the intent is to display
a patch, as git-log(1) says (and as it looks like).

(Note:  I'm willing to contribute to this task :-)

Cheers,
Charles

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: git log vs. svn log

Posted by Heikki Orsila <sh...@jolt.modeemi.cs.tut.fi>.
On Wed, Sep 05, 2007 at 08:00:38PM -0700, Karl Fogel wrote:
> shd@jolt.modeemi.cs.tut.fi (Heikki Orsila) writes:
> >> On 9/5/07, Heikki Orsila <sh...@jolt.modeemi.cs.tut.fi> wrote:
> >> > --stat is useful because I don't  want to write names of the
> >> > affected files into the changelog message.. Many people waste effort in
> >> > writing  those names, but the version control tool could
> >> > automagically do that.. And diffstat even shows how many lines changed..
> >> 
> >> 'svn log -v' already shows you the list of paths that changed in each
> >> revision.  I think that's what you want.  :-)
> >
> > To get a good overall picture of the change, I like to have the diffstat 
> > as well..
> 
> Want to file an issue pointing to this thread?  I'll bet someone picks
> it up pretty soon (though perhaps after 1.5, unless you have time to
> write the patch, which shouldn't be hard...)

Filed the issue.

-- 
Heikki Orsila			Barbie's law:
heikki.orsila@iki.fi		"Math is hard, let's go shopping!"
http://www.iki.fi/shd

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: git log vs. svn log

Posted by Karl Fogel <kf...@red-bean.com>.
shd@jolt.modeemi.cs.tut.fi (Heikki Orsila) writes:
>> On 9/5/07, Heikki Orsila <sh...@jolt.modeemi.cs.tut.fi> wrote:
>> > --stat is useful because I don't  want to write names of the
>> > affected files into the changelog message.. Many people waste effort in
>> > writing  those names, but the version control tool could
>> > automagically do that.. And diffstat even shows how many lines changed..
>> 
>> 'svn log -v' already shows you the list of paths that changed in each
>> revision.  I think that's what you want.  :-)
>
> To get a good overall picture of the change, I like to have the diffstat 
> as well..

Want to file an issue pointing to this thread?  I'll bet someone picks
it up pretty soon (though perhaps after 1.5, unless you have time to
write the patch, which shouldn't be hard...)

-Karl

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: git log vs. svn log

Posted by Heikki Orsila <sh...@jolt.modeemi.cs.tut.fi>.
On Wed, Sep 05, 2007 at 12:22:34PM -0500, Ben Collins-Sussman wrote:
> On 9/5/07, Heikki Orsila <sh...@jolt.modeemi.cs.tut.fi> wrote:
> 
> > --stat is useful because I don't  want to write names of the
> > affected files into the changelog message.. Many people waste effort in
> > writing  those names, but the version control tool could
> > automagically do that.. And diffstat even shows how many lines changed..
> 
> 'svn log -v' already shows you the list of paths that changed in each
> revision.  I think that's what you want.  :-)

To get a good overall picture of the change, I like to have the diffstat 
as well..

-- 
Heikki Orsila			Barbie's law:
heikki.orsila@iki.fi		"Math is hard, let's go shopping!"
http://www.iki.fi/shd

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: git log vs. svn log

Posted by Daniel Rall <dl...@collab.net>.
On Wed, 05 Sep 2007, Ben Collins-Sussman wrote:

> On 9/5/07, Heikki Orsila <sh...@jolt.modeemi.cs.tut.fi> wrote:
> 
> > --stat is useful because I don't  want to write names of the
> > affected files into the changelog message.. Many people waste effort in
> > writing  those names, but the version control tool could
> > automagically do that.. And diffstat even shows how many lines changed..
> 
> 'svn log -v' already shows you the list of paths that changed in each
> revision.  I think that's what you want.  :-)

And adding the names of the files changed yourself is useful if you
want to provide a detailed description of what was changed ('lsdiff
patchfile > msg' is helpful here).  Log messages which include symbol
names make hunting down bugs much easier, and symbol names make the
most sense when associated with file names...

Re: git log vs. svn log

Posted by Ben Collins-Sussman <su...@red-bean.com>.
On 9/5/07, Heikki Orsila <sh...@jolt.modeemi.cs.tut.fi> wrote:

> --stat is useful because I don't  want to write names of the
> affected files into the changelog message.. Many people waste effort in
> writing  those names, but the version control tool could
> automagically do that.. And diffstat even shows how many lines changed..

'svn log -v' already shows you the list of paths that changed in each
revision.  I think that's what you want.  :-)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: git log vs. svn log

Posted by Mattias Engdegård <ma...@virtutech.se>.
"Charles Acknin" <ch...@gmail.com> writes:

>Wouldn't 'svn diff | diffstat' do the job?

Probably not - this depends on what information you want. As I wrote earlier,
this will treat file deletions differently depending on whether their parent
directory is removed at the same time or not. It is also slow, especially for
tree copies (tags and branches).

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: git log vs. svn log

Posted by Charles Acknin <ch...@gmail.com>.
On 9/5/07, Heikki Orsila <sh...@jolt.modeemi.cs.tut.fi> wrote:
> On Wed, Sep 05, 2007 at 04:36:23PM +0100, Malcolm Rowe wrote:
> > On Wed, Sep 05, 2007 at 05:40:03PM +0300, Heikki Orsila wrote:
> > > $ git log -p
> > >
> > > commit 795688a4905257516886cac6f3add6ba467dfe0b
> > > Author: Heikki Orsila <sh...@thingol.localdomain>
> > > Date:   Wed Aug 22 23:42:24 2007 +0300
> > >
> > >     Add documentation of -i and -o [DOCUMENTATION]
> > >
> > > diff --git a/pmr.1 b/pmr.1
> > > index f42235a..d3fd6d2 100644
> > > --- a/pmr.1
> > > +++ b/pmr.1
> > > @@ -1,11 +1,11 @@
> > > -.TH pmr 1 "2006-11-03" Linux "user commands"
> > > +.TH pmr 1 "2007-08-22" Linux "user commands"
> > > ...
> > >
> >
> > I must admit, I'd find this useful as well.
>
> --stat is useful because I don't  want to write names of the
> affected files into the changelog message.. Many people waste effort in
> writing  those names, but the version control tool could
> automagically do that.. And diffstat even shows how many lines changed..

Wouldn't 'svn diff | diffstat' do the job?

Cheers,
Charles

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: git log vs. svn log

Posted by Heikki Orsila <sh...@jolt.modeemi.cs.tut.fi>.
On Wed, Sep 05, 2007 at 04:36:23PM +0100, Malcolm Rowe wrote:
> On Wed, Sep 05, 2007 at 05:40:03PM +0300, Heikki Orsila wrote:
> > $ git log -p
> > 
> > commit 795688a4905257516886cac6f3add6ba467dfe0b
> > Author: Heikki Orsila <sh...@thingol.localdomain>
> > Date:   Wed Aug 22 23:42:24 2007 +0300
> > 
> >     Add documentation of -i and -o [DOCUMENTATION]
> > 
> > diff --git a/pmr.1 b/pmr.1
> > index f42235a..d3fd6d2 100644
> > --- a/pmr.1
> > +++ b/pmr.1
> > @@ -1,11 +1,11 @@
> > -.TH pmr 1 "2006-11-03" Linux "user commands"
> > +.TH pmr 1 "2007-08-22" Linux "user commands"
> > ...
> > 
> 
> I must admit, I'd find this useful as well.

--stat is useful because I don't  want to write names of the 
affected files into the changelog message.. Many people waste effort in
writing  those names, but the version control tool could 
automagically do that.. And diffstat even shows how many lines changed..

-- 
Heikki Orsila			Barbie's law:
heikki.orsila@iki.fi		"Math is hard, let's go shopping!"
http://www.iki.fi/shd

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: git log vs. svn log

Posted by Malcolm Rowe <ma...@farside.org.uk>.
On Wed, Sep 05, 2007 at 05:40:03PM +0300, Heikki Orsila wrote:
> $ git log -p
> 
> commit 795688a4905257516886cac6f3add6ba467dfe0b
> Author: Heikki Orsila <sh...@thingol.localdomain>
> Date:   Wed Aug 22 23:42:24 2007 +0300
> 
>     Add documentation of -i and -o [DOCUMENTATION]
> 
> diff --git a/pmr.1 b/pmr.1
> index f42235a..d3fd6d2 100644
> --- a/pmr.1
> +++ b/pmr.1
> @@ -1,11 +1,11 @@
> -.TH pmr 1 "2006-11-03" Linux "user commands"
> +.TH pmr 1 "2007-08-22" Linux "user commands"
> ...
> 

I must admit, I'd find this useful as well.

Regards,
Malcolm