Re: Results of egrep -l '^<<<<<<< |^=======$|^>>>>>>> |^\|\|\|\|\\|\| '

[Paul Sander]

>--- Forwarded mail from Greg Woods:

>[ On Friday, July 20, 2001 at 11:40:08 (-0400), Noel L Yap wrote: ]
>> Subject: Re: Results of egrep -l '^<<<<<<< |^=======$|^>>>>>>> 
>> |^\|\|\|\|\\|\| '
>>
>> >You do nothing until you have to change the file in question.  Then
>> >before you commit it you do whatever's necessary syntactically to hide
>> >the markers.  That's it.  That's all.  I.e. you do the same thing you'd
>> >have to do with your own code, except you don't do it until/unless you
>> >have to change a given vendor file.  Pretty simple, eh?
>> 
>> If you do this, you're not really tracking the differences between what the
>> vendor gave you and what you've done to the file(s).  Instead, you're munging
>> two separate change sets into one checkin.

>Duh.  Do you expect the vendor to take all your changes no matter what?

No, but that's not what I'm arguing about.

I expect to be able to import the vendor's update, whatever it is, without
first having to make unrecorded changes to the code just to apease CVS.
Inserting escape characters, or otherwise quoting or hiding conflict markers
is such a change; THIS behavior must not be required before I commit the
vendor's code.

>Unless you make only the most basic changes (or you actually make them
>under the explicit instructions of the vendor), then you're *always*
>going to have local changes that are not taken up by the vendor!

The vendor typically makes an equivalent change that meets my needs, even
if it is not the exact change I sent.  In that case, I remove my local change
to bring my code closer to the vendor's.  Only occasionally do they reject my
change outright, in which case, yes, I diverge from the definitive sources.

>--- End of forwarded message from address@hidden