Re: Large patch/diff refuses to apply

On 10/06/16 09:19, Sean Bruno wrote:
> 
> 
> On 10/06/16 09:00, Alan Somers wrote:
>> On Thu, Oct 6, 2016 at 8:41 AM, Sean Bruno <sbruno_at_freebsd.org> wrote:
>>>
>>>
>>> On 10/06/16 08:23, Slawa Olhovchenkov wrote:
>>>> On Thu, Oct 06, 2016 at 07:57:57AM -0600, Sean Bruno wrote:
>>>>
>>>>> I'm doing a large amount of change to ixgbe(4) in support of IFLIB
>>>>> implementations and running into failures when trying to apply large
>>>>> diffs.  This is causing phabricator reviews to be unuseable as well.
>>>>>
>>>>> I've setup two trees to test this.  The first tree is used to generate
>>>>> the diff and the second (vanilla) is used to apply the diff.  The entire
>>>>> patch fails to apply, so I'm assuming that the size of the diff is
>>>>> failing because of a sanity check or something.
>>>>
>>>> No. This is expanded/collapsed keywords issuse:
>>>>
>>>> ===
>>>> -******************************************************************************/
>>>> -/*$FreeBSD$*/
>>>> ===
>>>>
>>>> svn diff over repo generate patch w/ collapsed keywords.
>>>> At working copy all keywords expanded.
>>>>
>>>
>>> Ah, I see.  Thank you.
>>>
>>> I am regenerating the failed files now.  That seems to work (if I leave
>>> the keywords alone).
>>>
>>> sean
>>>
>>
>> Also, I think the "svn patch" command (as opposed to plain "patch")
>> can deal with the RCS keywords.
>>
> 
> The use of "svn cp" also seems to generate a lot of failures as well.
> 
> I do want to maintain history between certain files when I go to commit.
>  e.g. svn cp fileA new_fileA, but this seems to confuse the crap out of
> diff/patch no matter what I do.  This leads to suffering with
> phabricator as well.
> 
> For now, I'm not going to do the "svn cp" part of my commit, but I will
> try and do this when I am ready to shove my work into current.
> 
> sean
> 

A good example of what I was trying to do:
svn mkdir sys/modules/ix/ix_iflib
svn cp sys/modules/ix/Makefile sys/modules/ix/ix/ix_iflib/Makefile
<edit sys/modules/ix/ix_iflib/Makefile>

That seems to really confuse things.

sean