[BLACKBOX] Aw: Re: [BLACKBOX] The official position of the Russian OberonCore team about BlackBox Support

From: Rene A. Krywult <"Rene>
Date: Fri, 23 Aug 2013 20:10:48 +0200

----boundary-LibPST-iamunique-2098560397_-_- Content-type: text/plain I agree that too much change is actually not the way to go. OTOH, the main power of OpenSource is that testing is done not by the core team, but by the community. Thus, there is no manpower needed for testing. This only works if you have frequent (whenever a single bug is fixed) updates on non-stable versions, while there are stable versions (once or twice a year at most) which are basically ready for production usage. Nevertheless, neither the frequent updates nor the stable versions should not change the whole system. As OberonCore said, the first objective is not to do harm. Also, I think the distinction between the basic framework and the components is essential. The basic framework are the components necessary for compiling and IDE. This includes database access and access to platform/operating system functions which are common to different systems. Those components need stable versions with a minimum of change. All other components may change as the author sees fit (though it is still stupid to change an existing interface without severe need). Rene Gesendet: Freitag, 23. August 2013 um 18:40 Uhr Von: "Wojtek Skulski" People use GNU/Linux systems because they follow "often changes, often > releases" scheme. I strongly disagree that it is a good method. In my experience it is terrible. Just to give an example: I adopted Analog Device Linux release 2.6.28 in 2009. It turns out that certain features were broken in that release. (Multicore does not work.) So I should upgrade. But certain other features were removed from later releases. (Namely, the Ethernet AX88180 driver.) So I am sticking to the 2009 release that I customized for my board that is using AX88180. I wrote extensive documentation for the customers how to use these Linux sources, how to compile, and which file is where in the source tree. Recently the embedded Linux community moved to the "buildroot" environment. It is absolutely, completely different from the previous one. The directories are different, the config files are in other places. Everything has been moved around. My current documents became completely useless. I will have to redo all my manuals from scratch, what is a month of work at least. What even worse, the Linux books describe kernel 2.6.x, while the buildroot uses kernel 3.x.x. So not only my documents are obsolete, but all the books that I have. And updated books have not been published yet. Given all the above, I am sticking with the sources from four years ago and I am delaying the transition to new sources. The fact that GNU/Linux have released several times since 2009 is of no relevance. My observations can be summarized as follows. The GNU/Linux way is only good for those folks who made GNU/Linux their way of life and their religion. Those folks are doing nothing in their life but following the recent greatest patches. They move files from place to place, change the file names, and they constantly change system interfaces. Those folks write books that are obsolete even before the books reach the bookstores. I have seen examples where a fresh book described subsystems that have been removed before the book was printed. If you are doing nothing else in your life but reading the Linux kernel mailing lists, then "often changes, often releases" scheme is for you. But if the computer is a tool rather than a substitute for the God, then OberonCore way is the right attitude. Thank you, Wojtek ---- To unsubscribe, send a message with body "SIGNOFF BLACKBOX" to LISTSERV{([at]})nowhere.xy ---- To unsubscribe, send a message with body "SIGNOFF BLACKBOX" to LISTSERV{([at]})nowhere.xy ----boundary-LibPST-iamunique-2098560397_-_- Content-type: text/html I agree that too much change is actually not the way to go.

OTOH, the main power of OpenSource is that testing is done not by the core team, but by the community. Thus, there is no manpower needed for testing.

This only works if you have frequent (whenever a single bug is fixed) updates on non-stable versions, while there are stable versions (once or twice a year at most) which are basically ready for production usage.

Nevertheless, neither the frequent updates nor the stable versions should not change the whole system. As OberonCore said, the first objective is not to do harm.

Also, I think the distinction between the basic framework and the components is essential.

The basic framework are the components necessary for compiling and IDE. This includes database access and access to platform/operating system functions which are common to different systems.

Those components need stable versions with a minimum of change.

All other components may change as the author sees fit (though it is still stupid to change an existing interface without severe need).

Rene

Gesendet: Freitag, 23. August 2013 um 18:40 Uhr
Von: "Wojtek Skulski" <skulski{([at]})nowhere.xy An: BLACKBOX{([at]})nowhere.xy Betreff: Re: [BLACKBOX] The official position of the Russian OberonCore team about BlackBox Support
> People use GNU/Linux systems because they follow "often changes, often
> releases" scheme.

I strongly disagree that it is a good method. In my experience it is
terrible.

Just to give an example: I adopted Analog Device Linux release 2.6.28 in
2009. It turns out that certain features were broken in that release.
(Multicore does not work.) So I should upgrade. But certain other features
were removed from later releases. (Namely, the Ethernet AX88180 driver.)
So I am sticking to the 2009 release that I customized for my board that
is using AX88180. I wrote extensive documentation for the customers how to
use these Linux sources, how to compile, and which file is where in the
source tree.

Recently the embedded Linux community moved to the "buildroot"
environment. It is absolutely, completely different from the previous one.
The directories are different, the config files are in other places.
Everything has been moved around. My current documents became completely
useless. I will have to redo all my manuals from scratch, what is a month
of work at least. What even worse, the Linux books describe kernel 2.6.x,
while the buildroot uses kernel 3.x.x. So not only my documents are
obsolete, but all the books that I have. And updated books have not been
published yet.

Given all the above, I am sticking with the sources from four years ago
and I am delaying the transition to new sources. The fact that GNU/Linux
have released several times since 2009 is of no relevance.

My observations can be summarized as follows. The GNU/Linux way is only
good for those folks who made GNU/Linux their way of life and their
religion. Those folks are doing nothing in their life but following the
recent greatest patches. They move files from place to place, change the
file names, and they constantly change system interfaces. Those folks
write books that are obsolete even before the books reach the bookstores.
I have seen examples where a fresh book described subsystems that have
been removed before the book was printed.

If you are doing nothing else in your life but reading the Linux kernel
mailing lists, then "often changes, often releases" scheme is for you. But
if the computer is a tool rather than a substitute for the God, then
OberonCore way is the right attitude.

Thank you,
Wojtek


----
To unsubscribe, send a message with body "SIGNOFF BLACKBOX" to LISTSERV{([at]})nowhere.xy

---- To unsubscribe, send a message with body "SIGNOFF BLACKBOX" to LISTSERV{([at]})nowhere.xy ----boundary-LibPST-iamunique-2098560397_-_--- Received on Fri Aug 23 2013 - 20:10:48 UTC

This archive was generated by hypermail 2.3.0 : Thu Sep 26 2013 - 06:29:50 UTC