get_source/hgforest.sh breaks "defpath -dgu"

Kelly O'Hair kelly.ohair at oracle.com
Tue Jun 26 18:13:58 UTC 2012


On Jun 21, 2012, at 6:54 PM, David Holmes wrote:

> On 22/06/2012 2:19 AM, Kelly O'Hair wrote:
>> The key is how you ran the initial clone command, did you do this:
>> 
>>     hg clone http://hg.openjdk.java.net/jdk8/tl   yourTL
>> 
>> OR
>> 
>>     hg clone http://hg.openjdk.java.net/jdk8/tl/  yourTL
> 
> The latter. It really shouldn't matter of course.
> 
>> I could change the make/scripts/hgforest.sh to not append / when the base path already has it.
> 
> BTW for the closed repos the // was in two places - again presumably because the "extra path" ended in /
> 
> Arguably it is a mercurial bug - it should ignore // in paths. :(

How about if I add some kind of filter on the paths I create, e.g.
      sed -e 's@\([^:]/\)//*@\1 at g'

???

-kto

> 
> Cheers,
> David
> 
>> -kto
>> 
>> 
>> On Jun 20, 2012, at 5:49 PM, David Holmes wrote:
>> 
>>> I just cloned the complete TL JDK8 forest by cloning the top repo then using get_source.sh to clone the other repos (open and closed).
>>> 
>>> At the top-level hgdefpath shows
>>> 
>>> /export/users/dh198349/jdk8>  hg defpath
>>>     default = http://hg.openjdk.java.net/jdk8/tl/
>>> default-push = None
>>> 
>>> I then went into the jdk repo to run "hg defpath -dgu dholmes" and got:
>>> 
>>> abort: http://hg.openjdk.java.net/jdk8/tl//jdk: Repository not found
>>> No hgrc files updated
>>> 
>>> Note the // prior to jdk. If I edit .hgrc/hgrc to change that to a / the defpath works.
>>> 
>>> Note that clones/pulls etc don't care about the // but the "defpath -dg" does.
>>> 
>>> All of the clones initiated via hgforest.sh have the // eg:
>>> 
>>> # Repos:  corba jaxp jaxws langtools jdk hotspot  jdk/src/closed jdk/make/closed jdk/test/closed hotspot/src/closed hotspot/test/closed deploy install sponsors pubs
>>> Starting on corba
>>> Starting on jaxp
>>> Starting on jaxws
>>> Starting on langtools
>>> # hg clone http://hg.openjdk.java.net/jdk8/tl//corba corba
>>> 
>>> David
>>> 
>> 




More information about the build-dev mailing list