QA Notice ...
 

With the latest portage I'm now getting.....

* QA Notice: //// double prefix

I know my setup is a little bit of a hybrid as my current EPREFIX is set
to "/", but I don't see any reason why this shouldn't work as it has
always done.

What happened ?

Alan.

QA Notice ...
 

On 23-12-2009 23:00:42 +0000, Alan Hourihane wrote:
> With the latest portage I'm now getting.....
>
> * QA Notice: //// double prefix
>
> I know my setup is a little bit of a hybrid as my current EPREFIX is set
> to "/", but I don't see any reason why this shouldn't work as it has
> always done.
>
> What happened ?

Not sure, but //// is pretty ugly I'd say so if we could reduce that to
a single / that would be nice.
Is this for all packages?


--
Fabian Groffen
Gentoo on a different level

QA Notice ...
 

On Thu, 2009-12-24 at 09:42 +0100, Fabian Groffen wrote:
> On 23-12-2009 23:00:42 +0000, Alan Hourihane wrote:
> > With the latest portage I'm now getting.....
> >
> > * QA Notice: //// double prefix
> >
> > I know my setup is a little bit of a hybrid as my current EPREFIX is set
> > to "/", but I don't see any reason why this shouldn't work as it has
> > always done.
> >
> > What happened ?
>
> Not sure, but //// is pretty ugly I'd say so if we could reduce that to
> a single / that would be nice.
> Is this for all packages?

Yes.

Alan.

QA Notice ...
 

On Thu, 2009-12-24 at 09:42 +0100, Fabian Groffen wrote:
> On 23-12-2009 23:00:42 +0000, Alan Hourihane wrote:
> > With the latest portage I'm now getting.....
> >
> > * QA Notice: //// double prefix
> >
> > I know my setup is a little bit of a hybrid as my current EPREFIX is set
> > to "/", but I don't see any reason why this shouldn't work as it has
> > always done.
> >
> > What happened ?
>
> Not sure, but //// is pretty ugly I'd say so if we could reduce that to
> a single / that would be nice.
> Is this for all packages?

That //// is coming from misc-functions.sh though. Because I guess ${ED}
is "/" then there's an additional "/", then ${EPREFIX} os "/" and
another "/" on the find command before the QA notice is issued.

Alan.

QA Notice ...
 

On Thu, 2009-12-24 at 10:01 +0000, Alan Hourihane wrote:
> On Thu, 2009-12-24 at 09:42 +0100, Fabian Groffen wrote:
> > On 23-12-2009 23:00:42 +0000, Alan Hourihane wrote:
> > > With the latest portage I'm now getting.....
> > >
> > > * QA Notice: //// double prefix
> > >
> > > I know my setup is a little bit of a hybrid as my current EPREFIX is set
> > > to "/", but I don't see any reason why this shouldn't work as it has
> > > always done.
> > >
> > > What happened ?
> >
> > Not sure, but //// is pretty ugly I'd say so if we could reduce that to
> > a single / that would be nice.
> > Is this for all packages?
>
> That //// is coming from misc-functions.sh though. Because I guess ${ED}
> is "/" then there's an additional "/", then ${EPREFIX} os "/" and
> another "/" on the find command before the QA notice is issued.

Actually this looks like ${ED} is broken as the packages are merging
to / instead of /var/tmp/portage/....

15027 works as expected so something changed since that one. I'll do a
diff to find the culprit.

Alan.

QA Notice ...
 

On Thu, 2009-12-24 at 20:58 +0000, Alan Hourihane wrote:
> On Thu, 2009-12-24 at 10:01 +0000, Alan Hourihane wrote:
> > On Thu, 2009-12-24 at 09:42 +0100, Fabian Groffen wrote:
> > > On 23-12-2009 23:00:42 +0000, Alan Hourihane wrote:
> > > > With the latest portage I'm now getting.....
> > > >
> > > > * QA Notice: //// double prefix
> > > >
> > > > I know my setup is a little bit of a hybrid as my current EPREFIX is set
> > > > to "/", but I don't see any reason why this shouldn't work as it has
> > > > always done.
> > > >
> > > > What happened ?
> > >
> > > Not sure, but //// is pretty ugly I'd say so if we could reduce that to
> > > a single / that would be nice.
> > > Is this for all packages?
> >
> > That //// is coming from misc-functions.sh though. Because I guess ${ED}
> > is "/" then there's an additional "/", then ${EPREFIX} os "/" and
> > another "/" on the find command before the QA notice is issued.
>
> Actually this looks like ${ED} is broken as the packages are merging
> to / instead of /var/tmp/portage/....
>
> 15027 works as expected so something changed since that one. I'll do a
> diff to find the culprit.

I've got the diff down to this at the moment which cures things for
me....

Any clues on this ?

Alan.

QA Notice ...
 

On 03-01-2010 21:48:00 +0000, Alan Hourihane wrote:
> > Actually this looks like ${ED} is broken as the packages are merging
> > to / instead of /var/tmp/portage/....
> >
> > 15027 works as expected so something changed since that one. I'll do a
> > diff to find the culprit.
>
> I've got the diff down to this at the moment which cures things for
> me....
>
> Any clues on this ?

Your diff basically has 1) STARTDIR/UNIXMODE filtering and 2) setting of
Prefix vars, which zmedico (Portage maintainer) believes are
unnecessary.

is 1) in any way relevant? I can imagine 2) to be related, since we
added it at some point as it was necessary.

Thanks for breaking this down.


--
Fabian Groffen
Gentoo on a different level

QA Notice ...
 

On Mon, 2010-01-04 at 17:10 +0100, Fabian Groffen wrote:
> On 03-01-2010 21:48:00 +0000, Alan Hourihane wrote:
> > > Actually this looks like ${ED} is broken as the packages are merging
> > > to / instead of /var/tmp/portage/....
> > >
> > > 15027 works as expected so something changed since that one. I'll do a
> > > diff to find the culprit.
> >
> > I've got the diff down to this at the moment which cures things for
> > me....
> >
> > Any clues on this ?
>
> Your diff basically has 1) STARTDIR/UNIXMODE filtering and 2) setting of
> Prefix vars, which zmedico (Portage maintainer) believes are
> unnecessary.
>
> is 1) in any way relevant? I can imagine 2) to be related, since we
> added it at some point as it was necessary.

UNIXMODE is for FreeMiNT, so that will need adding. But if I remove any
part of STARTDIR then things fall apart.

Also, I've tried removing the prefix var settings for EROOT and ROOT and
things also fall apart without it. I'll be trying EPREFIX when I get
chance.

Alan.

QA Notice ...
 

On 04-01-2010 16:26:26 +0000, Alan Hourihane wrote:
> > Your diff basically has 1) STARTDIR/UNIXMODE filtering and 2) setting of
> > Prefix vars, which zmedico (Portage maintainer) believes are
> > unnecessary.
> >
> > is 1) in any way relevant? I can imagine 2) to be related, since we
> > added it at some point as it was necessary.
>
> UNIXMODE is for FreeMiNT, so that will need adding. But if I remove any
> part of STARTDIR then things fall apart.
>
> Also, I've tried removing the prefix var settings for EROOT and ROOT and
> things also fall apart without it. I'll be trying EPREFIX when I get
> chance.

I'm analyzing this with the commits made here, and this is what I've
noticed sofar:


r15113 | grobian | 2009-12-18 09:22:04 +0100 (Fri, 18 Dec 2009) | 1 line

Remove redundant EPREFIX setting since it's already initialized earlier
in the config constructor. (r15044)

This is only the self["EPREFIX"] = EPREFIX for portageq bit.

I can't find the other in the separate commits, but I see it in the full
diff :S


--
Fabian Groffen
Gentoo on a different level

QA Notice ...
 

On Mon, 2010-01-04 at 18:01 +0100, Fabian Groffen wrote:
> On 04-01-2010 16:26:26 +0000, Alan Hourihane wrote:
> > > Your diff basically has 1) STARTDIR/UNIXMODE filtering and 2) setting of
> > > Prefix vars, which zmedico (Portage maintainer) believes are
> > > unnecessary.
> > >
> > > is 1) in any way relevant? I can imagine 2) to be related, since we
> > > added it at some point as it was necessary.
> >
> > UNIXMODE is for FreeMiNT, so that will need adding. But if I remove any
> > part of STARTDIR then things fall apart.
> >
> > Also, I've tried removing the prefix var settings for EROOT and ROOT and
> > things also fall apart without it. I'll be trying EPREFIX when I get
> > chance.
>
> I'm analyzing this with the commits made here, and this is what I've
> noticed sofar:
>
>
> r15113 | grobian | 2009-12-18 09:22:04 +0100 (Fri, 18 Dec 2009) | 1 line
>
> Remove redundant EPREFIX setting since it's already initialized earlier
> in the config constructor. (r15044)
>
> This is only the self["EPREFIX"] = EPREFIX for portageq bit.
>
> I can't find the other in the separate commits, but I see it in the full
> diff :S

Is there anything I can do to help what's needs to happen ?

At the moment, all I can see is that these changes need reverting to
unbreak things.

Alan.