Re: [gentoo-user] update fails, but I don't see why

[n952162 <n952162@web.de>]

On 12/13/20 10:02 PM, n952162 wrote:
> On 12/13/20 9:06 PM, Dale wrote:
>> Mark Knecht wrote:
>>>
>>> On Sun, Dec 13, 2020 at 12:03 PM n952162 <n952162@web.de
>>> <mailto:n952162@web.de>> wrote:
>>>> On 12/13/20 9:18 AM, Neil Bothwick wrote:
>>>>> Nearly 2 months, quite a long time in Gentoo update terms.
>>>>>
>>>>>
>>>> Okay, is the solution then to re-install?
>>>>
>>> Personally I wouldn't start with a reinstall.
>>>
>>> I'd start with my world file and remove/comment out every
>>> 'application' not required to keep the machine running/booting. That
>>> should be nearly everything other than things like grub, your kernels,
>>> hardware drivers, terminals, etc.
>>>
>>> NOTE: Just because you comment something out in the world file doesn't
>>> mean it won't run, it just won't get updated or be part of portage's
>>> considerations about what to do.
>>>
>>> At that point emerge @world should only be considering, from a build
>>> POV, the stuff you really need.
>>>
>>> This python problem everyone is having I cannot help with. Solving
>>> problems like that is why people run Gentoo. With lots of power comes
>>> lots of responsibility. However with all the applications
>>> 'unconsidered' you have a chance of getting the really important stuff
>>> built and booting.
>>>
>>> If I got that far then I'd start adding apps back in/uncommenting one
>>> or two at a time and see if I could make headway.
>>>
>>> I've updated machines that were a year out of date but it took days
>>> and days. If I didn't want to build code every week I decided I
>>> couldn't run Gentoo.
>>>
>>> HTH,
>>> Mark
>>
>> I agree.  I update once a week.  It seems a pretty good balance between
>> not having to do it to often and not having such drastic changes that it
>> makes things hard to work through.
>>
>> That said, when the tree is in the process of huge changes, it can
>> create problems even with weekly updates.  Right now, it is python and
>> the speed it is moving at.  Some versions that have been around for
>> ages, 2.7, is being removed.  Then python 3.6 is leaving etc etc etc.
>> Those of us that have been around long enough have seen this with other
>> packages as well.  Some packages are just hard to upgrade to begin with
>> and some create circular problems. The longer it goes between updates,
>> the larger that problem gets to be.  You get two different packages
>> doing that, you can find yourself running around in circles trying to
>> get emerge to chew what it can.
>>
>> While I usually do updates on Sunday evening, I'm considering doing
>> twice a week, Sunday and Wednesday.  At least until python settles down
>> a bit. Thing is, I think the worst part may be about over.  I think 2.7
>> is gone here, I think 3.6 is too.  That's two down.  It seems 3.7 and
>> above will be around a while but if they start going away soon, I may do
>> two updates a week if it starts making updates harder.
>>
>> Time can be a problem but sometimes it just depends on what packages
>> have changed and how fast they have changed.  For some systems that
>> haven't been updated in a while, having to remove python 2.7 and 3.6 in
>> one go, can cause problems that are hard to get around.  Right now just
>> isn't a good time to let updates get to far apart.
>>
>> The only thing that makes some of this survivable, getting help on this
>> mailing list.  Some people can decode the output of emerge and find a
>> way to work through it. Some are fairly easy, some not so much.
>> Sometimes removing/commenting out things in the world file will help.
>> Sometimes doing @system first helps. Sometimes you just have to update
>> certain packages in small chunks to get through a upgrade. Finding that
>> right option sometimes requires help.
>>
>> Just some thoughts.
>>
>> Dale
>>
>> :-)  :-)
>>
>>
>>
>
> My problem is I can't find a diagnostic methodology.  The one I most
> often hear is, update more often, or trail and error solutions.
>
> Does all that information in emerge's output point the way to the
> problem, and I just have to learn to understand it,  or am I just
> wasting my time there?  Are there better tools than log parsing? I know,
> there are lots of good tools,  but there needs to be methodology for
> using them (like understanding gentoo ;-) )
>
>

But commenting stuff out of the world file is at least a path. I'll
start with that tomorrow.