Re: heads-up: sh changes (perhaps fragility)

["Thomas Mueller" <mueller6725%twc.com@localhost>]

from Robert Elz:

    Date:        Mon, 05 Jun 2017 07:24:38 +0000
    From:        "Thomas Mueller" <mueller6725%twc.com@localhost>
    Message-ID:  <E2.00.03935.95705395@dnvrco-omsmta01>
        
  | Now I don't know which branch to stay with, lean toward staying with
  | current, but may wait until I hear better news about sh.
        
> There have (so far anyway, after the latest fixes) been no more reports
> about problems with sh.   Of course it has only been a few hours.

> I expect that the 2 fixes will be pulled up to netbsd-8 quite quickly,
> after which (until new stuff, rather than bug fixes) starts getting added
> in -current, the two should be the same.
        
> Pity that 8 missed the LINENO fixes. which will be next feature change
> to appear .. the changes that introduced the bugs came about because a
> rational $(( )) parsing/evaluating method was needed to be able to make
> LINENO expansion in there accurate... Saturday's changes were just a step
> along the way.
        
  | I was taken by surprise at the time of netbsd-7 branch when I thought
  | I was building NetBSD 6.99.46 or whatever, and I was updating to 7.99.1.
        
> That kind of change (and the recent 7.99.75 -> 8.99.1) are really almost
> no-ops, all that changes is the version string (and associated bookkeeping).

> kre

I guess I can feel free now to proceed with the update for amd64 and i386.

I am strongly inclined to stay with HEAD because of the prospect of getting earlier bug fixes (such as athn) and improvements.

UPDATE: Not so fast!  I tried the update on amd64 and found the resulting kernel broke dhcpcd and dhclient in combination with re(4) driver, making it impossible to access the Internet.

So I declined to install the userland (installworld), suppose this belongs in another thread; plan to update FreeBSD 11.0-STABLE as I wait on this matter.

Tom