[GTALUG] MP BIOS Toshiba - semi revival

[Christopher Browne]

On 17 March 2015 at 23:17, Anthony de Boer <adb at adb.ca> wrote:
> Lennart Sorensen wrote:
>> Recompiling serves no purpose.  Do you want to first recompile the
>> compiler before compiling the OS?  What are you going to compile the
>> compiler with?  How far down do you want to go?
>
> Too much of a modern system is in various scripting languages, which
> do effectively that everytime you run them.  Granted, so long as you're
> not doing tight inner loops in a script the performance hit isn't as
> bad as it can be.  Shell scripts that forked N things per loop iteration
> used to really crawl along, though the fact we had a couple of dozen
> users on a 386 running SVR3 might have had something to do with it too.
>
> Optimization is getting to be a lost art.
>
>> I think source distributions like gentoo are stupid, but at least they
>> only compile things once.  They aren't that crazy.  Some people like what
>> they can do with tweaking the settings and turning features on and off.
>> I prefer things that are tested and work and don't waste tons of CPU
>> time compiling what the distribution could have already compiled.
>
> Part of the reason I run Gentoo is to have the source code aboard my
> system and be sure the binaries were compiled from exactly that; their
> infrastructure facilitates that and the build process has to be robust
> enough to work on various strange folks' machines.  (The other part was
> at the time wanting something as unlike an RPM-based distro as possible
> due to having had enough of that for awhile.)
>
> Crazy would be taking something like Debian or Red Hat where you're
> supposed to love and run their distributed binaries and recompile them
> all yourself and find out how many builds only worked that once and are
> irreproduceable.  But someone somewhere has to keep them honest.  :-)

You don't need to compile everything *every* time to keep them honest;
you need to compile it *once*.  And it's not so much you as there being
*someone*.  And better still if the "someone" is an automated batch
process so that we can have a non-negligible amount of confidence
that it's repeatable.

And some of the "recompile for (imagined) security" takes this to
further heights of silliness...

- Do we need to recompile Bash (or Dash or zsh or whatever)
  each time we reboot?

- Oh dear, that means we need to recompile the Perl, Python, and
  Ruby distributions every time.  Should we be running the test
  suites, too, to verify that they're working as predicted?

- It seems idiotic to need to recompile KDE, libraries *and* apps.

- I'm running StumpWM as my window manager; this "security
  by recompiling everything" model means I need to recompile
  SBCL (the Common Lisp environment).

- Whoops, can we really trust things if we haven't recompiled
  GCC/LLVM since the last time we rebooted?  If recompiling
  code lends security, then surely not.

- Have you recompiled Grub lately?

And all of this falls out of deciding that when people say
"reliability," they don't *really* mean that; they really mean
"security."  And when they say "performance", they don't
*really* mean that; they really meant to say "security"
(even though they didn't, which ought to be a hint that
it wasn't what they meant).

Claim was made that Debian switched from using Bash
as the default shell (!= "default login shell", by the way)
"because security."  When the declared reasons didn't
have the word "security" anywhere.

But I guess that since *everything* is really computer
security, then the plans must be already well under way
for Debian to recompile everything, from the kernel to
Grub to all the scripting engines during the boot
process.
-- 
When confronted by a difficult problem, solve it by reducing it to the
question, "How would the Lone Ranger handle this?"