[rescue] Has freshmeat.net disappeared forever? WAS:::Re: What to do with the SS20

PAK pakenned-list at pobox.com
Wed Jun 25 09:23:24 CDT 2014


On Tue, Jun 24, 2014 at 04:42:39PM -0600, Richard wrote:
> In article <20140624220805.GA46307 at galah.local>,
>     PAK <pakenned-list at pobox.com> writes:
> 
> > Here's the joy of git--sometimes the effort needed to get something to
> > compile is simply "./configure && make && make install", [...]
> 
> ...which, by the way, is *not* portable.

I don't know whether you've missed my point, or just trying poke fun :)
I feel that you're somewhat deliberately truncating my argument.

But that's ok.  Let's take your point and use it to make mine :)

> ...which, by the way, is *not* portable.

Build processes in my sysadmin life were _not_ portable because we had a
fairly non-standard environment at the time (bleeding edge openLDAP,
PAM, and on and on... on both Solaris and Linux).  This meant that we
were writing patches to source code and authoring build and deployment
scripts.  All of that is non-portable (definitionally).  

For me, now, if I had to work in that sort of environment, it would be
_much_simpler_ with git.  It enables:

 *  Easily tracking my changes in version control (well, duh)
 *  Easily retrieving updated sources (trivially as compressed deltas,
    not another source download--git does this by design)
 *  Easily "replaying" my local changes to any (upstream) branch (git
    rebase)
 *  In case of conflicts, I can have git remember how I resolved
    conflicts before (git rerere)
 *  The integrity of the repository is cryptographically secure--no
    bitrot on flaky hardware.

So, as far as using git as a mechanism for distributing software (and
given what I know now), I would take it in a heartbeat.

PAK
    who is now stepping off his git soapbox...


More information about the rescue mailing list