On Nov 17, 2011, at 12:44 PM, Liam Proven wrote:
Which is why it's annoying that one of the
business arguments
/against/ FOSS is that either there is no support or that it's really
expensive.
Free community support is generally more use than the professional
paid-for stuff from the vendors, IMHO.
I agree wholeheartedly, with one caveat; you have to be willing to do at least some of the
work. The reason paid support is so popular in large companies and the government is
because both employ large numbers of people with no clue or motivation, so in order to get
*anything* fixed, you have to be able to just phone someone up and say, "I
power-cycled it and it's still not working, so please send me a new one no questions
asked and maybe you could reimage it for us too".
The converse, of course, is misc at
openbsd.org, which is the worst (or best, depending on
your point of view) place for clueless questions that amount to "do my work for
me". Things like "give me a step by step guide on how to make my random USB
touchpad work" get a swift and thorough lashing, and maybe deservedly so. But if
you're willing to at least try and figure out what your problem is, and work towards
finding a solution, FOSS communities are FAR more helpful than most scripted commercial
support.
Fun story: support isn't that much better for hardware manufacturers. We built an AMC
card using (among other things) a Freescale MPC8378 (PowerPC microcontroller) on board
that, at least for the first batch, had a few bizarre problems running at the promised 800
MHz. I traced it down to the fact that at some point, after heavy activity on the memory
controller at full speed, one or more of the byte lanes would be shifted by one
double-width (128-bit) word. It was a DDR2 controller, so obviously one of the
byte-synchronization FIFOs was getting thrown off by one tick, presumably due to a
poorly-constrained signal; we looked at the DDR2 signals coming in with a fast scope and
they looked beautiful, so we were pretty confident that wasn't it.
After being shunted around Freescale's Taiwan offices with no one saying anything
other than "well, you're probably not terminating it right, go re-spin your
board", they just stopped responding to us. Shortly thereafter, supplies of the
800MHz speed grade dried up for a few months; after shipments resumed, the problem was
fixed, so I guess we found a process/QA control problem for them? Th lack of
communication was just infuriating, though.
- Dave