16 Apr
2011
16 Apr
'11
9:48 a.m.
On Tue, Apr 12, 2011 at 05:06:23PM -0600, Richard wrote:
Yeah, so inscrutable that I was quickly able to add a feature (custom
message ids) to slrn (news reader) by writing a small patch for it. And
I'm definitely _not_ the worlds most awesome coder.
In article <C9CA264E.3E294%geoffr at zipcon.net>,
Geoffrey Reed <geoffr at zipcon.net> writes:
BTDT. One of many examples: HP-UX MCOE. The hardware (PA-RISC) is really nice.
The kernel is already something out of the bloody 80s. What do you mean,
you've got bugger all runtime tunables and twiddling most kernel parameters
needs a kernel relink and reboot[0]? On a HA environment (ok, thats what test
systems are for, but still). And don't get me started on the horrible crap
that HP ships as userland. The default cc is a crippled K&R ... thingy. Ok,
I get it, you are only supposed to use it for relinking the kernel and if
you are silly enough to want to compile some C code that was written in
the last 20 years, you are supposed to _buy_ the separate "ANSI C/C++
compiler". The system provided more binary is known to be broken since
basically forever (feed it something with long lines or even binary and it
will segfault). The system provided vi is a bad joke to anybody who spent
some time with less backwards Unix systems in the last 10 years. And I
have never seen such a horrible and convuluted hack to access ISO9660
filesystems (CDs/DVDs) before: IIRC (It's been a while) third party
software running in user space, talking to the block device, re-exporting
the contents as a special almost-but-not-quite-unlike local NFS server
that you can mount. *eeeeeekk*
Did I mention that the system tools are crap? The fastest way to unfuck
the HP-UX userland is to install the GNU toolchain, including gcc.
If it weren't for the nifty HA stuff in hardware and kernel as well as
for MC/ServiceGuard (which works quite well), it would be total crap.
And don't get me started on the (bloody expensive btw) commercial software
for it. With pearls like this one: A GUI that claims to be for X11 ... only
it was written by taking the Windows version and "porting it" (presumably,
On 4/12/11 3:16 PM, "Richard"
<legalize at xmission.com> wrote:
I think you have that backwards.
Either that or you *really* haven't spent much time looking at the
average open source project and only spent time looking at the good
ones you heard about by word of mouth. Like most things in the world, you get what you
pay for with open source.
Some open source is crap, but much of it isn't, [...] from the look of things, done by an overworked and
underskilled intern)
to X11 by compiling it against a library that emulates enough
of the Windows
build environment to get it to work on X11. The result is ... bloody damn
unusable and of course the application still thinks it is running on Windows.
And this is part of the reasons commercial, proprietary Unix is heading
for extinction.
At least with
open source software, if something doesn't work as you think
it should... You can fix it :)
Sorry, but I don't buy this either. Most open source software is
inscrutable to anyone but the original authors or current maintainers. Try fixing something in gcc on your own without having
worked on it
before, for instance.
Well, gcc is an interesting choice as by its very nature it is very, very
far from trivial. It currently supports 7 languages in the standard pack
(plus another 8 not in the default build) as well as 20 different cpu
architectures in the standard build (plus a ton more in different versions).
That is not exactly what I would choose as a starter project in open source
hacking.
Kind regards,
Alex.
[0] Admittedly, the HP-UX 11i for Itanium improved here: down to IIRC
less than 50% hard tunables (requiring kernel relink and reboot).
--
"Opportunity is missed by most people because it is dressed in overalls and
looks like work." -- Thomas A. Edison