On Thu, Jul 17, 2008 at 1:54 PM, Kenneth Marshall <ktm@rice.edu> wrote:
>> I think having the HASHVALUE_ONLY define is not a good idea -- it just
>> makes the patch harder to read. I suggest just removing the old code
>> and putting the new code in place. (That's why we have revision
>> control.)
>>
> One thing it helps is building an old version and a new version
> for comparative testing. Otherwise, you could end up with an apples-to-
> oranges comparison. I certainly think that the final patch should not
> have it, but it is useful now for testing and comparisons.
Yes, that's why Xiao did it that way. However, we traditionally just
submit a patch with only the changes and it's up to the person testing
to have an identical build-tree without the patch for testing.
Another reason for it is that even if you build without the define,
the patch author may have mistakenly added something outside the ifdef
which could impact testing.
I agree with Alvaro that we should submit it as a standard change patch.
--
Jonah H. Harris, Sr. Software Architect | phone: 732.331.1324
EnterpriseDB Corporation | fax: 732.331.1301
499 Thornall Street, 2nd Floor | jonah.harris@enterprisedb.com
Edison, NJ 08837 | http://www.enterprisedb.com/
