[gentoo-dev] gcc ebuild's, downgrades, deletion etc

[gentoo-dev] gcc ebuild's, downgrades, deletion etc

[Phil Richards]

Ok, I have sys-devel/gcc-3.2.2-r4 installed.
It was at some point unmasked.
I installed it.

My system is now trying to downgrade me to 3.2.2-r1 even though
I use --upgrade-only, and I'm guessing that the reason why is that
not only is 3.2.2-r3 onwards masked, but 3.2.2-r4.ebuild has been
deleted.  And, in fact, all reference to it in ChangeLog has gone.

*Please* *please* *please* don't do this.

If there is a problem with gcc-3.2.2-r4 by all means mask it;
preferably put some comment in the package.mask file *why* you are
masking it ("gcc with optimizations" doesn't tell me why it was
unmasked and then masked).

Then I can decide whether or not I should downgrade.

As it is I have no practical choice, nor any information on which to
make that choice.

As far as I can tell it is the deletion of the ebuild that forces
the downgrade - I am also concerned that a version that was visible
to ~x86 users has now been wiped from the world as if it never existed;
this leaves the possibility of a new, and different -r4 being released,
with no tracability of versions.

If nothing else, a comment in the ChangeLog explaining what was going
on would be good - that way you could have zapped the ebuild and people
would have had some idea what was going on.

And then I wouldn't have whinged, and the world would be a happier
place.  Even if only slightly.

phil

--
gentoo-dev@gentoo.org mailing list

RE: [gentoo-dev] Portage Integrity (Was: gcc ebuild's, downgrades, deletion etc)

[Todd Wright]

Phil Richards Wrote:

> My system is now trying to downgrade me to 3.2.2-r1 even though
> I use --upgrade-only, and I'm guessing that the reason why is that
> not only is 3.2.2-r3 onwards masked, but 3.2.2-r4.ebuild has been
> deleted.  And, in fact, all reference to it in ChangeLog has gone.
> 
> *Please* *please* *please* don't do this.
> 

I have to second this.  I recently submitted an ebuild (app-emulation/hercules-2.17.1.ebuild) and the developer apparently decided to copy the 2.15 ebuild to 2.17.1 rather than use the attached 2.17.1 ebuild provided.

Granted there were only minor changes in the new ebuild, but another more complex package may have seen more complex changes in the ebuild between the different versions, and added functionality is missing. He didnt even respond to my questions added to the 'bug', just updated one of the fields so I know he saw it.

I have to voice my concern for the integrity of the portage tree. If packages are being deleted and others being thrown together from an old copy instead of the fresh one, things are bound to break.

*Please* *please* *please* don't do this.
 
-- _--_|\ --------- Todd Wright -- wylie@geekasylum.org --------
  /      \            ICQ: 9589981   YIM: mvs38j
  \_.--._*  <---   http://www.geekasylum.org/~wylie/
        v       Mobile: +61-403-796-001    Ph: +61-2-9699-1746
----------------------------------------------------------------




--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Portage Integrity (Was: gcc ebuild's, downgrades, deletion etc)

[Spider]

[-- Attachment #1: Type: text/plain, Size: 1384 bytes --]

begin  quote
On Fri, 14 Mar 2003 11:47:05 +1100
"Todd Wright" <wylie@geekasylum.org> wrote:

> 
> I have to second this.  I recently submitted an ebuild
> (app-emulation/hercules-2.17.1.ebuild) and the developer apparently
> decided to copy the 2.15 ebuild to 2.17.1 rather than use the attached
> 2.17.1 ebuild provided.
>
> Granted there were only minor changes in the new ebuild, but another
> more complex package may have seen more complex changes in the ebuild
> between the different versions, and added functionality is missing. H


Now here is an issue and a pet peeve I have with ebuild submitters,
especially in the desktop set of applications.  They re-submit an ebuild
that is created by doing cp oldbuild->newbuild.  Why?

Doing that has made a lot of us cautious since the usersubmitted builds
are generally an annoyance in such cases, fex. When somone doesn't state
teh changes in the ebuild between the old and the new, I'm forced to
proofread the new builds completely as to avoid getting messed over.

So, did you actually state what differed your build from the last? 
builds without a Changes entry will get a far rougher treatment than
others.

(Why do users insist on not submiting the ChangeLog entry?) 

//Spider

--
begin  .signature
This is a .signature virus! Please copy me into your .signature!
See Microsoft KB Article Q265230 for more information.
end

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Portage Integrity (Was: gcc ebuild's, downgrades, deletion etc)

[Alain Penders]

On Fri, Mar 14, 2003 at 01:27:16PM +0100, Spider wrote:

> Now here is an issue and a pet peeve I have with ebuild submitters,
> especially in the desktop set of applications.  They re-submit an ebuild
> that is created by doing cp oldbuild->newbuild.  Why?
> 
> Doing that has made a lot of us cautious since the usersubmitted builds
> are generally an annoyance in such cases, fex. When somone doesn't state
> teh changes in the ebuild between the old and the new, I'm forced to
> proofread the new builds completely as to avoid getting messed over.
> 
> So, did you actually state what differed your build from the last? 
> builds without a Changes entry will get a far rougher treatment than
> others.
> 
> (Why do users insist on not submiting the ChangeLog entry?) 

Maybe because they trust that all our developers know how to diff a submitted
ebuild against the last approved one?

Even with a changelog entry, I would never add a user-submitted ebuild without 
diffing it and making sure I know what changed and why.

Alain

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Portage Integrity (Was: gcc ebuild's, downgrades, deletion etc)

[Spider]

[-- Attachment #1: Type: text/plain, Size: 1015 bytes --]

begin  quote
On Fri, 14 Mar 2003 06:19:17 -0700
Alain Penders <alain@gentoo.org> wrote:

> > I'm forced to proofread the new builds completely as to avoid
> > getting messed over.

> Maybe because they trust that all our developers know how to diff a
> submitted ebuild against the last approved one?

What do you think I mean with proofread, really?
(vimdiff and gtkdiff are both quite handy on larger stuff , gdm comes to
mind as more or less a nightmareish example) 

 
> Even with a changelog entry, I would never add a user-submitted ebuild
> without diffing it and making sure I know what changed and why.

of course not, I'm not inclined to have my or others systems compromised
or messed over, But my point is: adding a ChangeLog  or stating what is
done difference does make a change when submitting a build for something
thats already in the tree.


//Spider

--
begin  .signature
This is a .signature virus! Please copy me into your .signature!
See Microsoft KB Article Q265230 for more information.
end

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Portage Integrity (Was: gcc ebuild's, downgrades, deletion etc)

[Matt Tucker]

-- Spider <spider@gentoo.org> spake thusly:

> begin  quote
> On Fri, 14 Mar 2003 06:19:17 -0700
> Alain Penders <alain@gentoo.org> wrote:
> 
>> > I'm forced to proofread the new builds completely as to avoid
>> > getting messed over.
> 
>> Maybe because they trust that all our developers know how to diff a
>> submitted ebuild against the last approved one?
> 
> What do you think I mean with proofread, really?
> (vimdiff and gtkdiff are both quite handy on larger stuff , gdm comes
> to mind as more or less a nightmareish example) 
>  
>> Even with a changelog entry, I would never add a user-submitted
>> ebuild without diffing it and making sure I know what changed and
>> why.
> 
> of course not, I'm not inclined to have my or others systems
> compromised or messed over, But my point is: adding a ChangeLog  or
> stating what is done difference does make a change when submitting a
> build for something thats already in the tree.

While it's an excellent point that users should submit Changelogs, it
doesn't really address the original issue. To summarize, I believe the
conversation went like this:

user> I submitted a new ebuild with some changes, but the developer
      simply copied the old ebuild instead of using my new one.

dev> We have problems with users simply copying old builds to create
     the new one without submitting a Changelog. This makes it hard
     for us to figure out what's been changed.

I fail to see how this justifies using the old build instead of the new
one. I fail, in fact, to see how the reply is even related to initial
complaint.. If it's a problem for users to do "cp old.build new.build",
why is it okay for a dev to do it _when a new build has been supplied_?

And if there's not sufficient information to determine what's changed
in the ebuild (and you don't have the time to review a diff), wouldn't
it be better to kick it back to the submitter rather than ignoring what
they've submitted and using the old build for the new version? A diff
-q doesn't take any longer than a cp, and submitting changes without
submitting a changelog seems ample justification for refusing to commit
them.



--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Portage Integrity (Was: gcc ebuild's, downgrades, deletion etc)

[Jon Portnoy]

On Fri, Mar 14, 2003 at 03:12:04PM -0800, Matt Tucker wrote:
> -- Spider <spider@gentoo.org> spake thusly:
> 
> > begin  quote
> > On Fri, 14 Mar 2003 06:19:17 -0700
> > Alain Penders <alain@gentoo.org> wrote:
> > 
> >> > I'm forced to proofread the new builds completely as to avoid
> >> > getting messed over.
> > 
> >> Maybe because they trust that all our developers know how to diff a
> >> submitted ebuild against the last approved one?
> > 
> > What do you think I mean with proofread, really?
> > (vimdiff and gtkdiff are both quite handy on larger stuff , gdm comes
> > to mind as more or less a nightmareish example) 
> >  
> >> Even with a changelog entry, I would never add a user-submitted
> >> ebuild without diffing it and making sure I know what changed and
> >> why.
> > 
> > of course not, I'm not inclined to have my or others systems
> > compromised or messed over, But my point is: adding a ChangeLog  or
> > stating what is done difference does make a change when submitting a
> > build for something thats already in the tree.
> 
> While it's an excellent point that users should submit Changelogs, it
> doesn't really address the original issue. To summarize, I believe the
> conversation went like this:
> 
> user> I submitted a new ebuild with some changes, but the developer
>       simply copied the old ebuild instead of using my new one.
> 
> dev> We have problems with users simply copying old builds to create
>      the new one without submitting a Changelog. This makes it hard
>      for us to figure out what's been changed.
> 
> I fail to see how this justifies using the old build instead of the new
> one. I fail, in fact, to see how the reply is even related to initial
> complaint.. If it's a problem for users to do "cp old.build new.build",
> why is it okay for a dev to do it _when a new build has been supplied_?
> 
> And if there's not sufficient information to determine what's changed
> in the ebuild (and you don't have the time to review a diff), wouldn't
> it be better to kick it back to the submitter rather than ignoring what
> they've submitted and using the old build for the new version? A diff
> -q doesn't take any longer than a cp, and submitting changes without
> submitting a changelog seems ample justification for refusing to commit
> them.
> 
> 

My opinion here is: if you don't have the time to take a look at the 
diff, don't accept the bug containing the ebuild. Pass it on to someone 
who does have the time rather than risk borkage by just copying the old 
ebuild.

-- 
Jon Portnoy

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Portage Integrity (Was: gcc ebuild's, downgrades, deletion etc)

[Jon Portnoy]

On Fri, Mar 14, 2003 at 06:17:09PM -0500, Jon Portnoy wrote:
> On Fri, Mar 14, 2003 at 03:12:04PM -0800, Matt Tucker wrote:
> > -- Spider <spider@gentoo.org> spake thusly:
> > 
> > > begin  quote
> > > On Fri, 14 Mar 2003 06:19:17 -0700
> > > Alain Penders <alain@gentoo.org> wrote:

[snip]

> 
> My opinion here is: if you don't have the time to take a look at the 
> diff, don't accept the bug containing the ebuild. Pass it on to someone 
> who does have the time rather than risk borkage by just copying the old 
> ebuild.
> 

Just to add a few more thoughts: the user took the time to modify the 
old ebuild to work properly with the new version, test it, and submit it 
- a significant investment of time. To then not even take the time to do 
a diff and look over the changes is the same as saying "I don't value 
your contribution." That's not a good approach at all and certainly 
won't encourage people to help out.

-- 
Jon Portnoy

--
gentoo-dev@gentoo.org mailing list