[Pdx-pm] Fwd: [PLUG] ANNOUNCEMENT: Advanced Topics July 20th, 2005

[Eric Wilhelm]

I've mentioned this methodology to a few people in the group.  If you're 
interested, you may want to come to Jax tomorrow night and bash my 
casual "who needs testing?" approach (which is threatening to mature 
into the more conservative "who needs make install?".)  

Yes it's a linux group, but OS X has symlinks and subversion, so the 
mac-enabled can play too.

--Eric

----------  Forwarded Message:  ----------

Subject: [PLUG] ANNOUNCEMENT: Advanced Topics July 20th, 2005
Date: Tuesday 19 July 2005 02:53 pm
From: Alan <alan at clueserver.org>
To: plug-announce at lists.pdxlinux.org, PLUG list 
<plug at lists.pdxlinux.org>

Portland Linux/Unix Group Advanced Topics

Speaker:            Eric Wilhelm

Subject:            Rapid Development with a Safety Net

Date:               July 20th 2005
Time:               7:00pm - 9:00pm

Location:           Jax 826 SW 2nd Ave
                    Portland, OR

The talk will be on building multiple versions of software without
unexpected adverse difficulties. (And still have it work.)

The following is the outline as provided by Eric:

INFO

Audience:
  developers
  involved administrators
  consultants / integrators
  interested others

Description:
  The talk will follow the basic outline below, quickly covering
 typical version control layouts and usage.  Fairly thorough coverage
 of the perl/python library path structure (depending on audience
  familiarity.)

  I would expect the audience to bring some familiarity with system
  administration and program installation.  Developers will probably
  benefit more than administrators, but most admins have at least
  written a bit of perl code and would definitely benefit from a less
  ad-hoc approach to this (especially if they have trouble with
  different versions of custom scripts on multiple machines.)

  I would like to break into ad-hoc discussion about extending the
  concept into compiled code and large projects.  I think the audience
  would have more thoughts on handling dependencies and dealing with
  various package management systems.  These are somewhat hairy areas,
  and the conclusion may be that you only want to use this approach
 when the complexity is easily managed.

  I think discussion of a chroot approach would fit as well.

  The "stupid question threshold" would be down around an understanding
  of filesystems and code vs binary, so anyone with a few years
  experience and no fear of the command line should be able to keep up.


OUTLINE

Motivations:
  code development
  project tracking
  casual contribution

Traditional approach:
  make && make test && make install
  what sort of troubles copies cause
  why this is tedious if you're developing interpreted code
  why this is difficult if you're tracking multiple versions

Basics:
  version control (svn/cvs/svk)
    trunk, branch, etc.
    vendor branches, external
  symlinks
  locations
    /usr/local/svn_stuff/
    /usr/local/stow/
        etc.
    $PATH
    $PERL5LIB
        $PYTHONPATH

Interpreted code:
  perl, python, ruby, lisp

Compiled code:
  C, C++

Issues:
  Dependencies
  Package management systems
  broken symlinks
  unexpected behavior?


Usual meeting rules apply. Happy Hour meal prices for the first hour.


--
"All power is derived from the barrel of a GNU." - Mao Tse Stallman

_______________________________________________
PLUG mailing list
PLUG at lists.pdxlinux.org
http://lists.pdxlinux.org/mailman/listinfo/plug

-------------------------------------------------------

-- 
"Time flies like an arrow, but fruit flies like a banana."
--Groucho Marx
---------------------------------------------
    http://scratchcomputing.com
---------------------------------------------