Firefox is a PIG on memory var: Respectful critic - thanks.
Eric S. Johansson
esj at harvee.org
Wed Oct 26 15:47:28 UTC 2005
Art Alexion wrote:
> Nonsense. The strength of Open Source is its openness. Lots of talented
> people get to see the code, spot and diagnose problems and correct them
> quickly.
this is true only for a subset of open-source projects. In a posting I
made to private mailing list titled "the death of open-source universe
approaching", I pointed out a series of problems which effectively all
boil down to "there's no effect of way of paying someone so they will
notice your problem and help you fix it".
Most open-source projects are subsidized by an employer. The subsidy is
direct (i.e. Red Hat), or indirect (Joe sixpack Company employee). In
either case, free support can only be given to the limit of available
resources. In open-source projects I've been involved with, I need to
trade off development time versus support time. So if I spend my five
hours a week on support, no development happens and vice versa.
This support subsidy also limits the size a project can grow to. If the
project support requirements exceeds the capabilities of the primary
developers, then you have a higher proportion of ignored requests and
frustrated users. You might argue that other skilled users may be able
to step in but as I've seen on a couple of projects and have personally
experienced on IP cop, even with this willing group of support people,
the project can grow beyond their abilities to support.
A second factor is the expectation of free support. I tried selling
consulting services to support IP cop but gave up because of customer
expectations. Anyone who lives in an interrupt driven environment knows
that an interruption costs at least 15 minutes to get back to what they
were doing. Doctors know this, lawyers know this, accountants know this
and they all bill in minimum of 15 minute to 30 minute increments. Even
with explaining this, I still had people getting really upset that five
short phone calls cost them five 15 minute charges. After having to
deal with more than a couple of these "customers" who expect for fee
service to cost nothing, I gave up.
On the other hand, at times I've had the budget to pay someone for
support to help subsidize a project but it wasn't possible.
on the gripping hand, pay for support does not necessarily encourage
greater openness of solutions. Organizations like Red Hat etc. have an
economic incentive to not disclose problems and their solutions because
to do so removes a revenue source.
Quite the problem.
as to the assertion and there are lots of talented people looking at
code finding bugs, again only true for small number of projects. The
vast majority of projects I've been involved in have maybe three to five
people at best looking at problems because of the level of knowledge
needed to understand what's really going on. At best the average user
can do is give good bug reports.
personal example. in a past life, I worked with filesystems at a very
deep level and designed a remote procedure call based distributed file
system. When I run into problems with LVM2 or EVMS, I'm sure I could
figure out the problem at the cost of two weeks to a month of study.
The systems are not trivial. A simple change can screw your disks and
therefore, changes should be left up to the people with deep knowledge
(six months or more of bug fixing involvement). I'm not going to do
that because I have other itches to scratch.
fwiw, my personal projects from hell are samba, apache (destructive
complexity) LVM2, EVMS (insufficient documentation, diagnostic information)
> With proprietary software, only a few get to see the code, and
> non-monumental problems have to wait for "the next release" to get fixed
> -- and then it may get superceded by the needs of the marketing department.
and with open source, only a few understand the code. I think the real
difference is that with open source you get a chance at understanding
the code. It's up to you to decide whether or not it's an itch you need
to scratch or can sit back and ignore.
--- eric
More information about the ubuntu-users
mailing list