FAQ Search Today's Posts Mark Forums Read
» Video Reviews

» Linux Archive

Linux-archive is a website aiming to archive linux email lists and to make them easily accessible for linux users/developers.


» Sponsor

» Partners

» Sponsor

Go Back   Linux Archive > Gentoo > Gentoo User

 
 
LinkBack Thread Tools
 
Old 09-15-2008, 08:26 PM
"b.n."
 
Default I am a "f*****g retard". Can you help me?

Hi,

If someone if offended by the title, please don't. It's not my idea:
read below.


On the recent pointless, funny flame on "Portage VS Paludis", someone
linked this blog post on the usage of --resume-skipfirst:


http://r0bertz.blogspot.com/2007/01/be-careful-when-you-are-on-paludis.html

(read that and you will understand the title of my mail.)

I actually have to admit that, apparently, I suffer the same mental
disability of the poster -that is, I used to believe that using
--resume-skipfirst was harmless.


However in the post no actual explanation was done for that concerning
diagnosis.


Seriously: can someone more skilled than me explain why using
--resume-skipfirst and then trying to solve the unmerged packages is/can
be a bad idea? How can this break the system?


In return, I will bring flowers to Algernon.

Thanks a lot,
M.
 
Old 09-15-2008, 11:34 PM
Vaeth
 
Default I am a "f*****g retard". Can you help me?

> I actually have to admit that, apparently, I suffer the same mental
> disability of the poster -that is, I used to believe that using
> --resume-skipfirst was harmless.

In most cases it is harmless (e.g. if you use recompile your whole system
without installing anything new or upgrading anything).
Even in other cases a loop as mentiond on the webpage _can_ be used
safely - but this requires extreme care and/or a lot of knowledge.

The problem is that after failing of a package, portage does
not recalculate the dependencies, i.e. it will attempt to install also
those packages which depend on the failed package. Unfortunately, this
does not mean that the packages depending on this will fail either:
Due to the dubious autotools, they might recognize that the failed
package is not available (if it was a new packages which failed)
and build/install successfully but actually without support for this
package (although by the intention of the ebuild author and/or according
to your useflags they should support it).

The correct solution to the problem of failing compiles e.g. over night
is the proper implementation of the --keep-going option where these
problems do not arise, because dependencies are recalculated after a
failure. However, in the absence of this option, the above mentioned
loop was the _only_ available workaround. For the reason explained above
it _might_ give you an inconsistent system (although the chances for this
are not very likely) unless _you_ check/know the dependencies and if in
doubt recompile those packages which were "falsely" successfully installed
after the failed package.
So - as all workarounds - the above mentioned loop had pros and cons
(the pro is of course that, usually, your machine would compile through
the night, and if you are lucky and e.g. only one package fails and you
_know_ that nothing depends on it, you have not wasted the whole night
of compiling).
In the presence of a --keep-going option, it is now fortunately not
necessary anymore to weight the pros and cons. Of course, to insult
somebody just because he weighted the pros and cons differently is beyond
any acceptable limit.
 
Old 09-16-2008, 12:08 AM
Neil Bothwick
 
Default I am a "f*****g retard". Can you help me?

On Tue, 16 Sep 2008 01:34:31 +0200 (CEST), Vaeth wrote:

> The problem is that after failing of a package, portage does
> not recalculate the dependencies,

It does in 2.2.


--
Neil Bothwick

A. Top posters.
Q. What is the most annoying thing on Usenet?
 
Old 09-16-2008, 01:04 AM
"b.n."
 
Default I am a "f*****g retard". Can you help me?

Vaeth ha scritto:



I actually have to admit that, apparently, I suffer the same mental
disability of the poster -that is, I used to believe that using
--resume-skipfirst was harmless.



The problem is that after failing of a package, portage does
not recalculate the dependencies, i.e. it will attempt to install also
those packages which depend on the failed package.


OK, that's what flighted upon my head (Neil Botwick post notwithstanding)

I just assumed it recalculated dependencies and/or that there was some
mechanism immediately understanding that a necessary update is lacking.



to insult
somebody just because he weighted the pros and cons differently is beyond
any acceptable limit.


Of course -that's why the mail started being that sarcastic.

m.
 
Old 09-16-2008, 04:50 PM
Matthias Bethke
 
Default I am a "f*****g retard". Can you help me?

Hi b.n.,
on Mon, Sep 15, 2008 at 10:26:56PM +0200, you wrote:
> Seriously: can someone more skilled than me explain why using
> --resume-skipfirst and then trying to solve the unmerged packages is/can be
> a bad idea? How can this break the system?

Frankly I have no idea. I've heard that argument many times in the
Paludis discussions but never even an attempt at an explanation that
went beyond "it breaks your system". My understanding is that you can
have two kinds of situation if an upgrade fails:
a) the failed package is not a dependency of any other package
b) the failed package is a dependency of at least one other package
In case a) you get to keep the old version, no problem. In case b) the
package that depends on the failed one can
b1) work with the old version
b2) require the upgrade (and say so in the ebuild)
In case b1) things will continue working just fine. In case b2) you'll
get another failed emerge as portage will notice the unmet dependency,
so you get to keep the old version, no problem.

Did I miss anything? Sorry, no flowers today.

cheers,
Matthias
--
I prefer encrypted and signed messages. KeyID: FAC37665
Fingerprint: 8C16 3F0A A6FC DF0D 19B0 8DEF 48D9 1700 FAC3 7665
 
Old 09-16-2008, 04:57 PM
Matthias Bethke
 
Default I am a "f*****g retard". Can you help me?

Hi Vaeth,
on Tue, Sep 16, 2008 at 01:34:31AM +0200, you wrote:
> The problem is that after failing of a package, portage does
> not recalculate the dependencies, i.e. it will attempt to install also
> those packages which depend on the failed package.

OIC, so that was what I missed Somehow the thread got split up and I
missed your answer.

> In the presence of a --keep-going option, it is now fortunately not
> necessary anymore to weight the pros and cons. Of course, to insult
> somebody just because he weighted the pros and cons differently is beyond
> any acceptable limit.

++
I'd say "reimplement it properly" (i.e. check the deps) is always the
better approach than "the old implementation is b0rken so let's declare
the functionality so and not reimplement it at all".

cheers,
Matthias
--
I prefer encrypted and signed messages. KeyID: FAC37665
Fingerprint: 8C16 3F0A A6FC DF0D 19B0 8DEF 48D9 1700 FAC3 7665
 
Old 09-16-2008, 05:00 PM
Neil Bothwick
 
Default I am a "f*****g retard". Can you help me?

On Tue, 16 Sep 2008 18:50:35 +0200, Matthias Bethke wrote:

> > Seriously: can someone more skilled than me explain why using
> > --resume-skipfirst and then trying to solve the unmerged packages
> > is/can be a bad idea? How can this break the system?
>
> Frankly I have no idea. I've heard that argument many times in the
> Paludis discussions but never even an attempt at an explanation that
> went beyond "it breaks your system". My understanding is that you can
> have two kinds of situation if an upgrade fails:
> a) the failed package is not a dependency of any other package
> b) the failed package is a dependency of at least one other package
> In case a) you get to keep the old version, no problem. In case b) the
> package that depends on the failed one can
> b1) work with the old version
> b2) require the upgrade (and say so in the ebuild)
> In case b1) things will continue working just fine. In case b2) you'll
> get another failed emerge as portage will notice the unmet dependency,
> so you get to keep the old version, no problem.
>
> Did I miss anything?

I think you missed an important part of the Gentoo philosophy, that it
gives you the loaded gun but it's up to you to not point it at your foot.
Not providing options that could potentially break a system in certain
circumstances is for a Nanny Distro. Here the ethos is "here's the tool,
read the man page and don't blame us if you do something stupid".

Does paludis also refuse to unmerge packages in the system set?


--
Neil Bothwick

Eagles may soar, but Wombles don't get sucked into jet engines
 
Old 09-16-2008, 06:32 PM
Alan McKinnon
 
Default I am a "f*****g retard". Can you help me?

On Tuesday 16 September 2008 19:00:33 Neil Bothwick wrote:
> I think you missed an important part of the Gentoo philosophy, that it
> gives you the loaded gun but it's up to you to not point it at your foot.
> Not providing options that could potentially break a system in certain
> circumstances is for a Nanny Distro. Here the ethos is "here's the tool,
> read the man page and don't blame us if you do something stupid".
>
> Does paludis also refuse to unmerge packages in the system set?

I like the traditional behaviour of portage. When an update fails it tends to
say:

"You asked me to do something. It didn't work; here's the output. Have a look
at it then tell me what to do next. I'm a dumb piece of software, you are the
thinking human so don't expect me to think for you."

A failed emerge is by definition an error, and unpredictable. How can we
expect software to dream up the best solution to an exception?

--
alan dot mckinnon at gmail dot com
 
Old 09-16-2008, 06:43 PM
Dale
 
Default I am a "f*****g retard". Can you help me?

Alan McKinnon wrote:

On Tuesday 16 September 2008 19:00:33 Neil Bothwick wrote:


I think you missed an important part of the Gentoo philosophy, that it
gives you the loaded gun but it's up to you to not point it at your foot.
Not providing options that could potentially break a system in certain
circumstances is for a Nanny Distro. Here the ethos is "here's the tool,
read the man page and don't blame us if you do something stupid".

Does paludis also refuse to unmerge packages in the system set?



I like the traditional behaviour of portage. When an update fails it tends to
say:


"You asked me to do something. It didn't work; here's the output. Have a look
at it then tell me what to do next. I'm a dumb piece of software, you are the
thinking human so don't expect me to think for you."


A failed emerge is by definition an error, and unpredictable. How can we
expect software to dream up the best solution to an exception?





I have used --skipfirst before but I also don't think it is a good
idea. For a idiot like me to say that must mean something. I guess it
would depend on the package as to whether it could be skipped or not.
If I do a emerge -e world, I prefer nothing to fail else what is the point?


Later.

Dale

:-) :-)
 
Old 09-16-2008, 07:04 PM
Vaeth
 
Default I am a "f*****g retard". Can you help me?

Alan McKinnon wrote:

> "You asked me to do something. It didn't work

But it is an annoyance if you leave your computer on during the three
days you are on the road to compile a load of new packages like e.g.
a new kde version, and when you return, compiling has not even started
because your first pacakge fortune-mod had a trivial file collision.

> A failed emerge is by definition an error, and unpredictable. How can we
> expect software to dream up the best solution to an exception?

Since the best solution to this exception is to finish that part of
the task which is not influenced by this error, I think the
expectation for this exception is clear.
 

Thread Tools




All times are GMT. The time now is 04:42 PM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.
Copyright 2007 - 2008, www.linux-archive.org