RFR 8005962 : TEST_BUG: java/util/Properties/MacJNUEncoding can fail in certain environments

Stuart Marks stuart.marks at oracle.com
Fri Jan 11 20:35:36 UTC 2013

On 1/11/13 9:57 AM, Brent Christian wrote:
> On 1/11/13 9:54 AM, Chris Hegarty wrote:
>> On 11/01/2013 17:50, Brent Christian wrote:
>>> I did an 'hg mv' of the directory containing the (two) test files. But I
>>> then had to 'hg export' to send the changeset for pushing. I'm guessing
>> Yes, I noticed this before. 'hg export' and 'hg mv' are not friends.
>> -Chris.
> I see that hg also has a 'bundle' and 'unbundle'.  Has anyone used it? Would
> that have been better in this situation?

Aha, so it was 'hg export' that was the culprit. I should have figured that 
since you (Brent) authored the changeset but Naoto had pushed it.

In any case, "hg export --git" will export the changeset in a fashion that 
preserves the rename history. This should work fine with "hg import". As far as 
I know, the --git option isn't the default because other patch-importing 
programs (such as "patch" itself) don't understand it.

Bundles will also preserve rename information. I was a fan of bundles for a 
while. They're more difficult to work with, though. First, they're binary. 
Second, if you don't have the parent changeset of what's in the bundle, it's 
totally opaque; you can't see what's inside of it.

I'd recommend using "hg export --git".


More information about the core-libs-dev mailing list