IT bubble days are over. Linux kernel (almost the only useful FOSS thing left) tries to not
loose it's volume.
Developers do what they do on limited resources, using dumb tools or no tools at all, saving
buzz-hype using PR fuzz. It is a social and economical problem, not a technical and/or ethical
one.
If GRSecurity/PaX wants to save this only FOSS thing from security holes, then systematic
education, automation of audit and code creation by tools must be employed, database of broken
C/API usage must be collected, those patterns must be coded and automated by static analysis
on the source editing stage.
This must big and hard step, step forward. But your even smaller resources are wasted to fix
and confront consequences. Also i don't see something mentioned in your workflow/goals. And
investing into something as moot as all this even for "better future", isn't the way business
wants to make money today.
Top developers try to stay in tune with hardware support/performance side of the technology
development. Or they do, what they know how to do best or just somehow (another fs, another
scheduler, etc.). They and many others are doing this by basic think+edit+patch+debug(rc1,
rcX)... No `git` or similar can eliminate "thinking" and knowledge there.
While automated checking and automated creating of code may reduce debug stages and predict
problems on level higher than `gcc -Wall`. Again step forward. But this again against those
companies, who milk from static analysis, security audit and consulting.
Linus tries to bring tools, but all so far made, were NIH reimplementation of existing ones.
Systematic approach like regression tracking is a craft of committed individuals, no automated
tools are available also. And their work is also in some kind of fuzzy state.
Problem of those cargo-cult security specialists, grown under the shadows of LKML and dark
ages of Linux 0.99, now being on the top of the food chain and don't want to listen or
revolutioning something, is one of all human kind. Just look at present science and thus
education, politics, economics, etc... Unless they will gain more, one have no luck. In case
of retire, make sure you/your agenda have educated and skilled youngsters to propose and
compete.
Posted Jun 17, 2008 19:36 UTC (Tue) by aleXXX (subscriber, #2742)
[Link]
> IT bubble days are over. Linux kernel (almost the only useful FOSS
> thing left) tries to not loose it's volume.
Ok, so let's ignore gcc, Apache, Eclipse, bind, sendmail, postfix, ruby,
python, cmake, eCos, WebKit/KHTML, Firefox, OpenOffice.org, KDE, Gnome, X
Windows, LLVM, php, Tomcat, Samba, FreeBSD, Darwin, sdcc, mplayer, ...
Alex
"Stable" kernel 2.6.25.7 released
Posted Jun 17, 2008 23:36 UTC (Tue) by olecom (guest, #42886)
[Link]
Exactly!
Seriously:
* binutils/as with really good macro support using `sed`
* zoo of web servers (one's problem if Java version x.y.z.X.Y.Z is needed somewhere)
* Emacs (though nothing more that just text editing is useful in any "IDE")
* bind && sendmail are pride of security holes of past and present (there are alternatives
like exim and qmail for the latter, and some others for former)
* don't know any useful application of zoo of scripting languages except flame wars, otoh i'd
like to have shell be not as it was 20 years ago. Now python 3000 is going to be Java-like,
Ruby runs out of gas (afaik).
* cmake? Oh, man: shell (with one Linux and one libc it's not an autoconf-like mess any more),
ccache (which is 2 time faster shell + sed anyway)
* eCos? maybe, as well as zoo of other abandoned small/tiny OSes
* web browsers? lynx + nc + sed (quite seriously, even for some needed javascript)
* office? in case of bureaucracy maybe, FYI there's only MS there
* X and other useless stuff for what? for global warming?
* php, Tomcat, sdcc (same as above)
* Samba (see office)
* FreeBSD != FOSS, it's a BSD, LSD and such
* Darwin. Don't know, can't say anything
+ mplayer? yes, unless google will open all-codecs-and-file_formats-->ogg
and btw it runs on in pure text-mode using memory mapped framebuffer of modern VGAs just fine
i'd add xpdf (yea, needs X), free console fonts (terminus), inn2, slrn (nothing is hype, but
are very useful).
more importantly accessibility of PDF and web content to simple ASCII + images representations
and no JavaScript/Flash. No w3c handwaving, but real value of information with high SNR.
In case if you are an engineer in architecture, electronics, etc., you will have expensive
proprietary CAD systems, possibly based on Linux kernel and some version of GNU libc. Real
industry is not a joke, handcrafting or religion. Eclipse platform and derivatives doesn't
belong here or they are fancy solutions around basic functionality.
"Stable" kernel 2.6.25.7 released
Posted Jun 18, 2008 6:48 UTC (Wed) by aleXXX (subscriber, #2742)
[Link]
Could you please try to be consistent at least with yourself ?
Alex
"Stable" kernel 2.6.25.7 released
Posted Jun 18, 2008 13:41 UTC (Wed) by olecom (guest, #42886)
[Link]
>> IT bubble days are over. Linux kernel (almost the only useful FOSS
>> thing left) tries to not loose it's volume.
>
>Ok, so let's ignore gcc, Apache,
Or simply "almost the only critical (functionality and security) OS part, useful FOSS thing
with no alternatives" to pass flamewar guard, and let us on the subject and >>the message
body<<.