Is the long error message at rev-upgrade wanted or a bug?

Joshua Root jmr at macports.org
Fri Apr 5 03:46:17 PDT 2013


On 2013-4-5 19:55 , Clemens Lang wrote:
> On Fri, Apr 05, 2013 at 07:47:35AM +1100, Joshua Root wrote:
>> I don't think revupgrade_scanandrebuild should itself be throwing an
>> error because building a port failed. Or if it does, the error should
>> be caught by the caller.
>>
>> Changing: error "<message>"
>> to: ui_error "<message>"; return 0
>> would give the desired behaviour, I think.
> 
> I'd favor adding code catching the error at the caller level, e.g. in
> macports::revupgrade in macports1.0/macports.tcl. There's a semantic
> difference between "revupgrade returned 0 and previously printed some
> text" and "an error occured and revupgrade stopped". In case of an
> error, macports::revupgrade should probably return != 0, because that's
> what port/port.tcl checks for as far as I can see (and it also prints
> the ticket URL, if it failed).

Yeah, that does make sense, since the revupgrade_scanandrebuild return
value is used for something else.

- Josh


More information about the macports-dev mailing list