[23888] trunk/base/src/port1.0/portlivecheck.tcl

[Weissmann Markus]

On 12 Apr 2007, at 15:05, Paul Guyot wrote:

> On Apr 12, 2007, at 8:54 PM, source_changes at macosforge.org wrote:
>
>> Revision: 23888
>>           http://trac.macosforge.org/projects/macports/changeset/ 
>> 23888
>> Author:   eridius at macports.org
>> Date:     2007-04-12 04:54:10 -0700 (Thu, 12 Apr 2007)
>>
>> Log Message:
>> -----------
>> Fix livecheck to check for master_sites properly.
>> If a portfile isn't set up for livecheck and isn't a sourceforge/ 
>> freshmeat project, default to none, not freshmeat (which would  
>> cause a failure).
>> Clean up some if checks to avoid unnecessary syntax
>
> and
>
>> Revision: 23889
>>           http://trac.macosforge.org/projects/macports/changeset/ 
>> 23889
>> Author:   eridius at macports.org
>> Date:     2007-04-12 04:55:42 -0700 (Thu, 12 Apr 2007)
>>
>> Log Message:
>> -----------
>> Make the up-to-date message for livecheck be info, not debug
>
> Hello Kevin,
>
> As I understand it, in addition to fixing and cleaning the code,  
> you changed the semantics of livecheck significantly (without  
> modifying the documentation).
> Previously, livecheck would:
> - report an error if it didn't work
> - default to freshmeat if nothing was set
> - be silent if the port was up to date
>
> Now, livecheck will:
> - report an error if it didn't work
> - be silent if nothing was set
> - report if the port was up to date
>
> The old observed behavior when everything was ok (nothing was  
> displayed) now means that nothing was checked, which is quite a  
> problem if the goal of a maintainer is to keep their ports up to  
> date. The advantage of the original semantics is that it drives  
> port maintainers to write livecheck for their ports, while setting  
> a good default so that most ports don't need a specific livecheck.  
> Basically, you livecheck your ports and any output means that you  
> have to do something, shall it implement a specific livecheck  
> because default doesn't work or update because the port was updated  
> upstream.
>
> What argument do you have in favor of this change of semantics? Can  
> we decide on reverting to the previous behavior?
>

I'll have to chime in on this one: I also prefer to (i) get  
maintainers to add livecheck and (ii) get information if I have to  
actually do something.


-Markus

---
Markus W. Weissmann
http://www.mweissmann.de/