[oi-dev] pkg5 question

John Center john.center at villanova.edu
Thu Jan 12 17:10:56 UTC 2012


Hi Andrzej,

Thanks for replying, this came in the nick of time!  I added an #ifdef 
with the strndup function to _actions.c just to get past this problem & 
then ran up against the various *at functions that S10 is missing.  I'll 
pull this down & take a look.

	-John


On 01/12/2012 12:04 PM, Andrzej Szeszo wrote:
> Hi John
>
> It is possible that the current version of pkg5 is relying on the
> strndup now (yes S11 and illumos both have this function now). For
> s10-userland we are using a bit older version of pkg5 from March 2011
> and have not attempted to update it to the current one yet.
>
> We don't see strndup problem on Solaris 10 probably because we use an
> older codebase. Additionally we apply some patches which you can find
> here:
> <http://hg.openindiana.org/users/aszeszo/s10-userland/file/90aab7c315bf/components/pkg5/patches>.
> Some of the patches are environment specific. Some of the were required
> to get things working on Solaris 10 in general.
>
> Here are the details the pkg5 from s10-userland is at:
>
> changeset:   2251:00ccbd44fcbc
> tag:         tip
> user:        Padraig O'Briain<padraig.obriain at oracle.com>
> date:        Fri Mar 11 13:38:36 2011 +0000
> summary:     17213 Typos in man page for packagemanager and pm-updatemanager
>
> You can run 'hg update 00ccbd44fcbc' to switch to this version.
>
> Andrzej
>
>
> On 01/12/12 02:06 PM, John Center wrote:
>> On 01/11/2012 04:05 PM, John Center wrote:
>>> _actions.so calls strndup(), which doesn't exist on S10.  (Does S11 have
>>> it now?)  I can add a strndup.c to fix this, but I'm not sure how to do
>>> it.  The pkg gate source doesn't use makefiles for the modules
>>> directory, it looks like setup.py does the compiling&   linking itself,
>>> so I'm not sure what gets added where.  If anyone has a clue how to add
>>> this in&   have it built, I'd greatly appreciate the help.  Otherwise,
>>> I'll have to paste the function into _actions.c.
>>>
>> I made an assumption here that this would be the preferred way to fix
>> this.  Am I wrong?  I do plan to submit my changes upstream, once I
>> get something working.
>>
>> Thanks.
>>
>>      -John
>>
>>




More information about the oi-dev mailing list