[OE-core] Bug: 5089

[Bruce Ashfield]

On Fri, Aug 30, 2013 at 9:02 AM, Laszlo Papp <lpapp at kde.org> wrote:
> On Fri, Aug 30, 2013 at 1:47 PM, Bruce Ashfield <bruce.ashfield at gmail.com>
> wrote:
>>
>> On Fri, Aug 30, 2013 at 7:45 AM, Laszlo Papp <lpapp at kde.org> wrote:
>> > Hi,
>> >
>> > there was this change which also goes undocumented. Unfortunately, this
>> > is
>> > causing a lot of headache for me, but I expect that for others, too.
>> >
>> > I would like to propose the default kernel and module name from:
>> >
>> > uImage--3.2.1...
>> >
>> > to:
>> >
>> > uImage-3.2.1...
>> >
>> > This would avoid a lot of work for people. I have been trying to fix
>> > issues
>> > for one day due to this which is unfortunate.
>> >
>> > Also, the double dash on its looks silly IMO. It should be either '-0-'
>> > or
>> > '-'. I prefer the latter for compatibility reasons, although someone
>> > could
>> > argue about that the former is more readable.
>>
>> Having maintained several systems when the default name was changed
>> to include PE, it didn't cause any problems here. Both dependent recipes
>> and scripts picked up the change easily and convenience symlinks made
>> the -- or -0- not something that needed to be dealt with.
>
>
> That makes me think you have too flexible scripts which will cause issues
> elsewhere due to the flexibility. I had been on that front, too. That was a
> gate in my case for a lot of issues where something was expected, but the
> script was too flexible to verify properly. Either, when you use an accurate
> script, i.e. you follow the name format specified, it will cause you issues
> when updated. It is likely that you just have not provided accurate enough
> pattern matching.
>

Everyone's workflow is different, and I'm confident that the infrastructure
I use around the build is solid .. it's chugged along for over three
years now :)
But I don't disagree that output and format changes impact everyone differently.

Cheers,

Bruce

> Anyway, I will not insist on this much if people think breaking this for
> many scripts is fine, and extending the documentation should be enough.
>
> -- Laszlo



-- 
"Thou shalt not follow the NULL pointer, for chaos and madness await
thee at its end"