bugsquad
Thanks John ! Thanks Luis, and happy birthday to you
and have fun ! (IIRC Flat Top is that pool place close to MIT,
and I enjoyed it a lot too)
libxml2 and libxslt
More releases, more bug found, more bug fixed. Yesterday
I cleaned up the use of _private field in the libraries, so
to should ease the life of the people using it for wrappers
like PHP and Perl bindings. I also wanted to check that the
recent changes didn't break the very large file support, this
still works fine (RHEL AS 3, Celeron, 4.8 GByte XML file):
server:~/XML -> ls -l db24000000.xml
-rw-rw-r-- 1 veillard vcsa 4843680040 Feb 27 15:57 db24000000.xml
server:~/XML -> xmllint --stream --timing db24000000.xml
Parsing took 1153361 ms
server:~/XML -> xmllint --stream --timing --relaxng db.rng db24000000.xml
Compiling the schemas took 1 ms
Parsing and validating took 2348820 ms
db24000000.xml validates
server:~/XML -> ./testSAX --timing db24000000.xml
768000006 callbacks generated
Parsing took 468294 ms
server:~/XML -> grep -C 2 MHz /proc/cpuinfo
model name : Celeron (Coppermine)
stepping : 6
cpu MHz : 701.620
cache size : 128 KB
fdiv_bug : no
server:~/XML ->
OSS job
lauris, you're right, it's hard to get
paid for OSS jobs, the industry is built only around the
fact that they are gonna pay only if they are forced to.
I think it's the lesson I take from Red Hat Linux end of life (i.e.
the company had to force corporate users to buy the
support -- well that's my own analysis of the Red Hat
Enterprise Linux/Fedora
change), I think it's unfortunate that short vision on
investment is the common rule. It doesn't mean that OSS
can't feed you, but it's tricky. Things like GPL'ing and
offering dual licences seems the most common way to reach
the corporate wallet. Your point about fame vs. income is
partly right, releasing under a liberal licence also allow
to grow mindshare, then the potential customer base is quite
larger, seems to me the two approachs possible are really:
- Strict GPL licencing, smaller mindshare but allows
you to sell under a dual licence, this is fits
better if you target a smaller specific
market.
- Very liberal licencing (Apache/BSD/MIT) allows you
to get a larger mindshare, but selling your
services is harder, it's better suited for very
common pieces if you can get a big piece of the
"market", Apache is a prime example.
About paying for support, I'm still a bit dubious about
this model, in the sense I see it as requiring a lot of
resources to be viable, a large organization can make an
economy of scale for servicing similary a large number of
users, but for a small business based on OSS, seems to me
you better off trying to keep the number of bugs as small
as possible, limit the time spent dealing with trivial
issues and instead try to cash on new feature requests,
optimization or special development which sounds a better
use of the time and deep knowledge of the code which are the
developer main assets.
Trying to live of OSS in a small structure still looks
extremely difficult to me, even if the corporate world
start to understand what OSS and Free Software are.
