Re: svn commit: r33117 - in branches/ignore-mergeinfo/subversion: libsvn_client tests/cmdline

[Julian Foad <ju...@btopenworld.com>]

On Wed, 2008-09-17 at 08:26 -0500, Hyrum K. Wright wrote:
> Daniel Shahaf wrote:
> >> On the ignore-mergeinfo branch:
> >> Teach the client diff APIs to honor the --ignore-mergeinfo flag.  This
> >> does not yet implement this for 'diff --summarize', so we error out
> >> for that condition.
> > 
> > Short(er) option for this?  Maybe --no-g alias?
> 
> I'm really ambivalent about this; we can bikeshed about it later, 

+1!

> I'd rather
> just get the feature written so we can decide whether to merge it or not.

Hyrum, I was talking to Paul about changing the mergeinfo handling to
not add/change mergeinfo on nodes that didn't actually change. He thinks
it's feasible. If we can do that, there will then be much less need for
this option. That would be the best thing in the world, I think.

However, even then, it would still be useful, especially on "svn diff"
where the aim is to review changes or send a patch to someone, and on
"svn log" of historical data; it's just that "svn status" should be nice
and clean anyway, so we might want to omit the command-line option from
"status". Note that "status" is where the biggest concern was over the
negative consequences of hiding this information - the concern that
people would look at "status" and then commit only the items shown as
modified.

This change to mergeinfo isn't certain, but I hope it works.

- Julian



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

Re: svn commit: r33117 - in branches/ignore-mergeinfo/subversion: libsvn_client tests/cmdline

["Hyrum K. Wright" <hy...@mail.utexas.edu>]

Julian Foad wrote:
> On Wed, 2008-09-17 at 08:26 -0500, Hyrum K. Wright wrote:
>> Daniel Shahaf wrote:
>>>> On the ignore-mergeinfo branch:
>>>> Teach the client diff APIs to honor the --ignore-mergeinfo flag.  This
>>>> does not yet implement this for 'diff --summarize', so we error out
>>>> for that condition.
>>> Short(er) option for this?  Maybe --no-g alias?
>> I'm really ambivalent about this; we can bikeshed about it later, 
> 
> +1!
> 
>> I'd rather
>> just get the feature written so we can decide whether to merge it or not.
> 
> Hyrum, I was talking to Paul about changing the mergeinfo handling to
> not add/change mergeinfo on nodes that didn't actually change. He thinks
> it's feasible. If we can do that, there will then be much less need for
> this option. That would be the best thing in the world, I think.
> 
> However, even then, it would still be useful, especially on "svn diff"
> where the aim is to review changes or send a patch to someone, and on
> "svn log" of historical data; it's just that "svn status" should be nice
> and clean anyway, so we might want to omit the command-line option from
> "status". Note that "status" is where the biggest concern was over the
> negative consequences of hiding this information - the concern that
> people would look at "status" and then commit only the items shown as
> modified.

Agreed.  I would much rather see us not generate the noise to begin with than to
have to mask it later on.  I won't proceed any farther on 'svn st -u' until this
discussion has reached some conclusion.

> This change to mergeinfo isn't certain, but I hope it works.

So do I!

-Hyrum

Re: svn commit: r33117 - in branches/ignore-mergeinfo/subversion: libsvn_client tests/cmdline

[Paul Burba <pt...@gmail.com>]

On Wed, Sep 17, 2008 at 10:21 AM, Julian Foad
<ju...@btopenworld.com> wrote:
> On Wed, 2008-09-17 at 08:26 -0500, Hyrum K. Wright wrote:
>> Daniel Shahaf wrote:
>> >> On the ignore-mergeinfo branch:
>> >> Teach the client diff APIs to honor the --ignore-mergeinfo flag.  This
>> >> does not yet implement this for 'diff --summarize', so we error out
>> >> for that condition.
>> >
>> > Short(er) option for this?  Maybe --no-g alias?
>>
>> I'm really ambivalent about this; we can bikeshed about it later,
>
> +1!
>
>> I'd rather
>> just get the feature written so we can decide whether to merge it or not.
>
> Hyrum, I was talking to Paul about changing the mergeinfo handling to
> not add/change mergeinfo on nodes that didn't actually change. He thinks
> it's feasible. If we can do that, there will then be much less need for
> this option. That would be the best thing in the world, I think.
>
> However, even then, it would still be useful, especially on "svn diff"
> where the aim is to review changes or send a patch to someone, and on
> "svn log" of historical data; it's just that "svn status" should be nice
> and clean anyway, so we might want to omit the command-line option from
> "status". Note that "status" is where the biggest concern was over the
> negative consequences of hiding this information - the concern that
> people would look at "status" and then commit only the items shown as
> modified.
>
> This change to mergeinfo isn't certain, but I hope it works.

Hyrum,

FYI: The thread this is being discussed is
http://svn.haxx.se/dev/archive-2008-09/0443.shtml.

Paul

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