SCIENTIFIC-LINUX-USERS Archives

May 2021

SCIENTIFIC-LINUX-USERS@LISTSERV.FNAL.GOV

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Yasha Karant <[log in to unmask]>
Reply To:
Yasha Karant <[log in to unmask]>
Date:
Wed, 5 May 2021 15:36:28 -0700
Content-Type:
text/plain
Parts/Attachments:
text/plain (49 lines)
What you describe -- replacing a distro's utilities by those from other 
than the distro -- is done in practice for *SOME* things, as most on 
this list do.  However, under no condition should this be called a 
stable distro, let alone an "enterprise hardened stable" distro, without 
the amount of testing that most of us on this list realistically cannot 
provide.

I can provide examples.

Clearly, such "one-offs" are NOT production distros as you point out.

The real problem comes about when system core packages, such as "glib", 
Python, etc, in which the stock distro has a particular release level 
woven throughout much of the system core -- again, a "one-off" distro, 
but with all sorts of instabilities or even total failure.

Depending upon the access level any of the above has to other parts of a 
system, there could be large security vulnerabilities introduced.


On 5/5/21 3:21 PM, Konstantin Olchanski wrote:
> On Tue, May 04, 2021 at 11:00:00PM +0100, Andrew C Aitchison wrote:
>> On Tue, 4 May 2021, Konstantin Olchanski wrote:
>>
>>> - first slide of "distribution landscape" is nonsense,
>>> with everybody stuck with el7 for another 3 years and
>>> bye, bye, c++14, c++17, c++20.
>>
>> Is Red Hat Developer Toolset 10
>> https://urldefense.proofpoint.com/v2/url?u=https-3A__access.redhat.com_documentation_en-2Dus_red-5Fhat-5Fdeveloper-5Ftoolset_10_html_10.0-5Frelease-5Fnotes_dts10.0-5Frelease-23Changes-5Fin-5FDTS&d=DwIBAg&c=gRgGjJ3BkIsb5y6s49QqsA&r=gd8BzeSQcySVxr0gDWSEbN-P-pgDXkdyCtaMqdCgPPdW1cyL5RIpaIYrCn8C5x2A&m=Chv7ZrLnQII3dvv1o48KTg7YAxT9SNUUCkOkflNOCd4&s=bZUlWtCI03IJ95QWY-_DTkmqMdwOzXq8PVBkiR1dZLM&e=
>> not an option ?
>>
>> (OK, C++20 support in g++ 10.2.1 is "experimental).
>>
> 
> And so what?
> 
> I can take SL-6 and graft modern versions of all important packages,
> one does not even need the devtoolset, GCC is easy to build from sources.
> 
> But this is no longer "SL-6", it is "SL-6-KO1", at best.
> 
> Same thing, "CentOS-7 with devtoolset, php from webtatic, python from pip, kernel from ELREPO, etc" is not CentOS-7.
> 
> It is an irreproducible Franken-monster-bashed-together-locally thing.
> 
> Is this the new standard, the best way to go, "the new thing" for production environments?
> 

ATOM RSS1 RSS2