On Tue, 29 Dec 2009, Jonathan Rochkind wrote:
I think you may find yourself somewhat in the minority in thinking Apache is
bad software. (I certainly have my complaints about it, but in general I find
it more robust, flexible, and bug-free than just about any other software I
work with).
But aside from getting into a war about some particular package: It may be
true that in general popular software does not necessarily equal good
software -- even popular open source software. And doesn't neccesarily equal
the right software solution for your problem. (I could mention some
library-sector-origin open source software I think proves that, but I won't,
and it would just be my opinion anyways, like yours of Apache).
But popular software _does_ mean software that has a much higher chance of
continuing to evolve with the times instead of stagnating, getting it's bugs
and security flaws fixed in a timely manner, and having a much larger base of
question-answering and support available for it (both free and paid).
Which is one important criteria for evaluating open source software. But
nobody was suggesting it should be the _only_ criteria used for evaluating
open source software, or even neccesarily the most important. It depends on
your situation.
I think that part of the problem here is that software tends to fill a
niche, and some of these larger software projects tend to fill the
'enterprise' niche.
Now, Apache 2 in many ways *is* easier to configure than Apache 1.3, but
the sheer number of configuration options from all of the different
modules makes it more difficult to configure than the Netscape/iPlanet/
SunOne product line. (at least to me, other people might not be making
the sorts of changes that I deal with).
However, there's a lot of power in Apache's configuration ability ... I
just wish I didn't have to deal with all of it.*
... but it's like anything -- if I switch to a different server, it might
be easier to configure, but then I lose mod_perl support, so it's a
trade-off.
-Joe
* I think I lost a week trying to get some software virtual hosts working
correctly, where there'd be a 'default' host, and one that only
responded to specific names and had some alternate security options.
