[pkg-discuss] [REVIEW] Fixes for bugs 6342, 7712, 8109, 8896, 8897, 8898

[Shawn Walker]

Stephen Hahn wrote:
> * Shawn Walker <swalker at opensolaris.org> [2009-05-13 22:20]:
>> Greetings,
>>
>> The following webrev contains changes for the following bugs and RFEs:
>>
>>   6342 pkgrecv should support local disk repository
>>   7712 pkgrecv stack traces when (very) incomplete FMRI specified
>>   8109 status during package download can exceed 80 characters
>>   8896 traceback when opening server manifest using CachedManifest
>>   8897 pkg.publish.transaction close should not fail with 500 if depot  
>> close fails
>>   8898 pkg.publish.transaction open timestamp support desired
>>
>> webrev:
>> http://cr.opensolaris.org/~swalker/pkg-6342/
...
> 
>   man/pkgrecv.1.txt:
...
>   105.  Was PKG_SRC_REPO previously documented?  PKG_DEST and
> 	PKG_SRC_REPO should match.  (Or is PKG_SRC_REPO not taking a
> 	file path a permanent state?)

Previously, no environment variables were implemented (supported) or 
documented for pkgrecv.  You're correct that eventually pkgrecv might 
eventually accept a file path, so it seems right to change it to PKG_SRC.

>   modules/server/transaction.py:
> 
>   153ff.  (Future-oriented nit.)  You may want to leave an XXX comment,
> 	as publishing with a modified timestamp will require some kind
> 	of publication authorization above the basic "can open
> 	transactions" authorization.

I didn't think it was too much of an issue since they can't republish 
packages already in the catalog this way (it will fail on close), and 
they can't modify existing packages this way.  Although, I was thinking 
about adding a "reopen" operation for that...

>   pull.py:
> 
>    59.  Please file a bug for a pkg.command module.  error() should be
> 	in a common module for our commands.  And abort().  (Unless
> 	there's a common Python module that has better implementations.)

Filed 8946.

-- 
Shawn Walker