18-Apr-2024 18:08 GMT.		 UNDER CONSTRUCTION
Anonymous, there are 69 items in your selection [1 - 50] [51 - 69]
[News] New Hans-Joerg Frieden interviewANN.lu
Posted on 14-May-2002 18:50 GMT by Jon69 comments
View flat
View list
Read interview here. And Please: no flame war!
New Hans-Joerg Frieden interview : Comment 1 of 69ANN.lu
Posted by Jon on 14-May-2002 16:53 GMT
Oh no, the link disappeared :(
Once again:
http://www.student.oulu.fi/~jniemima/interviews/hans-joerg_frieden.html
New Hans-Joerg Frieden interview : Comment 2 of 69ANN.lu
Posted by Ole-Egil Hvitmyren on 14-May-2002 17:08 GMT
It's always good to get a "fix". Ahhhh. I really enjoy these little snippets of info, I just wish there were more!!! I need another shot!!! Mind you, some days it feels like an overdose...
New Hans-Joerg Frieden interview : Comment 3 of 69ANN.lu
Posted by cOrpse on 14-May-2002 17:14 GMT
In reply to Comment 2 (Ole-Egil Hvitmyren):
Yep its good to have something which isn't a negative comment for once . Whats happened to everyone ? some of the amiga peeps are acting like their girlfriend/boyfriend , family and fav. musical artist all got killed in a pile up.
I hear os4 in coming with a needle and powder combo to get the downers over it :P
New Hans-Joerg Frieden interview : Comment 4 of 69ANN.lu
Posted by José on 14-May-2002 17:17 GMT
One ignorant question: won't the libraries be somewhat slower with the new functionality added to them?
New Hans-Joerg Frieden interview : Comment 5 of 69ANN.lu
Posted by cOrpse on 14-May-2002 17:21 GMT
In reply to Comment 4 (José):
Its running on a faster cpu/system , so it shouldn't really degrade the experience , it just won't be as fast as it would have been on the new system with out improvement..
I'd rather have to extra functions tho :).
New Hans-Joerg Frieden interview : Comment 6 of 69ANN.lu
Posted by Ben on 14-May-2002 17:40 GMT
In reply to Comment 4 (José):
Expect serious speed - even emulation is x10 faster on average for most things...
New Hans-Joerg Frieden interview : Comment 7 of 69ANN.lu
Posted by William F. Maddock on 14-May-2002 18:04 GMT
In reply to Comment 4 (José):
Once loaded into RAM, it will only be slower if it's churning through more code to do the same things as before (without even factoring in the system speed-up ='D ). It's easy to forget that just because code gets loaded doesn't mean it gets executed. If it doesn't hit the processor it doesn't slow you down.
New Hans-Joerg Frieden interview : Comment 8 of 69ANN.lu
Posted by Anonymous on 14-May-2002 18:07 GMT
In reply to Comment 4 (José):
>One ignorant question: won't the libraries be somewhat slower
>with the new functionality added to them?
No, the number of functions contained in a library has absolutely no bearing on speed.
New Hans-Joerg Frieden interview : Comment 9 of 69ANN.lu
Posted by smithy on 14-May-2002 20:29 GMT
>Some things have been changed dramatically. For example, the libraries have
>moved ahead from the old concept of a single rigid jump table.
I am concerned as to why they've designed a new library format. What was wrong with the jumptable model? Or is it the case they are now using the ELF library format (they are using GCC?).
New Hans-Joerg Frieden interview : Comment 10 of 69ANN.lu
Posted by [JC] on 14-May-2002 21:15 GMT
In reply to Comment 9 (smithy):
The jumptable model is a pain in the arse compared to symbol based linkage. That was one of the things I found I much preferred when i moved from the amiga to win32
New Hans-Joerg Frieden interview : Comment 11 of 69ANN.lu
Posted by tinman on 14-May-2002 21:57 GMT
In reply to Comment 10 ([JC]):
Why is it such a pain in the ass? And why would a regular programmer even have to involve themselves in what structure a library is, unless you were creating header files/pragmas/MSequivalent for your compiler (chances are it would be provided with the library anyway, or there would be a tool to do it automatically)?
New Hans-Joerg Frieden interview : Comment 12 of 69ANN.lu
Posted by Anonymous on 14-May-2002 23:10 GMT
Facts we can glean from this interview:
# Hyperion aren't eating their own dog food yet (*)
# Hyperion have no previous experience with this work
I don't think there's much hope of OS4 reaching end users this summer.
(*) For those who haven't done software development, "eating your own dog food" is a phrase Microsoft uses for the practice of making programmers actually use the software they plan to foist on users. If the people making the dog food won't eat it, what makes you think the dogs will? Eating dog food is 100% essential if you're going to ship an OS that works more than half the time.
New Hans-Joerg Frieden interview : Comment 13 of 69ANN.lu
Posted by bhickman on 14-May-2002 23:47 GMT
In reply to Comment 12 (Anonymous):
Pardon me while I pull my foot out of your arse... but where are those facts? How come all of a sudden one person is now all of Hyperion? And please, tell us why they would be using the system when it's not 100% completed. Also, I recall that Hyperion has previous experience working with the development of OSX on the Mac, so they do have prior experience in the area of developing an OS.
To quote the original post: "And Please: no flame war!"
Glass houses man. Don't live in one. Hell, I understand why the Friedens don't read the forums now. All these false facts flying around like this.
-Bill
New Hans-Joerg Frieden interview : Comment 14 of 69ANN.lu
Posted by [JC] on 15-May-2002 00:06 GMT
In reply to Comment 11 (tinman):
With LVO's, you have to assign them an offset, which effectively allocates space in a jump table. With symbol based linkage, you mark the symbol as exported, and that's it. It's easy, it's fast, and what's more there's no looking up what LVO does what.
Furthermore, it would be in Amiga's interest to follow a common EXE format like ELF. This increases the chance of people developing tools for it, and also means existing tools can be ported relatively easily.
New Hans-Joerg Frieden interview : Comment 15 of 69ANN.lu
Posted by kevin orme on 15-May-2002 04:35 GMT
In reply to Comment 12 (Anonymous):
one essential point to remember - if you were lucky enough to attend Carl Sassenrath's slide show at AmiWest about 3 years ago, you would remember photos of RJ and CS working with HP workstations that ran 'virtual' Amigas, because Jay M and others were still building the hardware (remember the joe pillow story?). Didn't seem to affect their ability to make computer history in a negative way that I can see.
In other words, don't be so quick to jump to conclusions about the developers not having 'actual' Amigas to work on - OS 1.x turned out pretty revolutionary anyway, and provided a good foundation for the rest of the incremental and major improvements that came later, which of course WERE done on Amigas by that time, as you can see in Dave Haynie's deathbed vigil video.
If you can't have confidence in those doing the work, *give up* and use a Winblows box (and wait for your monthly round of service packs - the fact that OS3.9 only had 2 is pretty impressive in my mind - NT had >6 SP's because there was a TCP stack bug in SP6 which thus required SP6a immediately afterward, [funny that a bunch of bugfixes were required to fix previous bugfixes] and Win2K is already past 2SPs, XP is about to have its first already). Patience is a virtue, which of course has been sorely tested by the years since Commodore's demise, to be sure. But maybe this time we'll be rewarded - don't you want to be there to find out? If not we can then bitch for many years about 'what was' anyway and fill Christian's 'unmoderated' forum for many years to come, right? :)
kevin orme
amiga university
www.amigau.com
New Hans-Joerg Frieden interview : Comment 16 of 69ANN.lu
Posted by Akaru on 15-May-2002 05:11 GMT
In reply to Comment 12 (Anonymous):
/*
(*) For those who haven't done software development, "eating your own dog food" is a phrase Microsoft uses for the practice of making programmers actually use the software they plan to foist on users. If the people making the dog food won't eat it, what makes you think the dogs will? Eating dog food is 100% essential if you're going to ship an OS that works more than half the time.
*/
Geez if Microsoft actually lived by those rules they'd have a decent product, as it is they're selling some rank ass dog food that is 90% chalk, and 10% horse uhhh....
Well you get the idea.
New Hans-Joerg Frieden interview : Comment 17 of 69ANN.lu
Posted by dakang on 15-May-2002 05:28 GMT
In reply to Comment 12 (Anonymous):
Why are you here? Why dont you go off and try to get XP working or download a Linux ISO or something?
Your negative comments are not warranted. You're expecting OS4.0 to be coded by a team of 100 on big salaries - thats not the case....
New Hans-Joerg Frieden interview : Comment 18 of 69ANN.lu
Posted by DaveW on 15-May-2002 06:14 GMT
In reply to Comment 14 ([JC]):
In intruigued as to what he meant by inlining the old functions ( hopefully on an as called basis rather than inlining the whole lot ). Sure this gets past the fact that architecturally the library has changed on the other hand its going to increase the size of recompiled classic applications software some ( whilst increasing performance ) and it means that if classic function is ever upgraded or even patched the app needs to be recompiled to bring it in.
Perhaps I misread what he put.
I agree that the old library architecture was a painus in the an*s.
Dave.
New Hans-Joerg Frieden interview : Comment 19 of 69ANN.lu
Posted by Jon on 15-May-2002 06:19 GMT
In reply to Comment 12 (Anonymous):
Dear Anonymous,
Even when I try to stay objective...I have to disagree a bit.
Well, it might be true that the core team (I mean the actual Hyperion people) hasn't made an operating system before but I guess that nobody is questionalizing their skills and knowledge. Then again the 3.x developers might have worked for years with their piece of software.
Dogfood part: I believe that people making specific component (PPC filesystem, TCP/IP stack) use them very intensivily already (or if you don't, please inform us :)
New Hans-Joerg Frieden interview : Comment 20 of 69ANN.lu
Posted by Jon on 15-May-2002 06:24 GMT
Hmm.. btw if eating dog food is essential I guess I never be an os developer then ;)
I know dogs who eat everything..
New Hans-Joerg Frieden interview : Comment 21 of 69ANN.lu
Posted by Robert "RAG" Gustavsson on 15-May-2002 09:07 GMT
In reply to Comment 12 (Anonymous):
Ok, then that is why they do the Windows development on FreeBSD machines, run Hotmail on Solaris etc?
RAG
New Hans-Joerg Frieden interview : Comment 22 of 69ANN.lu
Posted by Phill on 15-May-2002 09:24 GMT
In reply to Comment 15 (kevin orme):
> NT had >6 SP's because there was a TCP stack bug in SP6 which thus required
> SP6a immediately afterward, [funny that a bunch of bugfixes were required to
> fix previous bugfixes] and Win2K is already past 2SPs, XP is about to have
> its first already).
WinNT/2000/XP have to be stable as they are used in workstation/server environments where you can't have down time etc. Windows NT 4 was current for approximately five years, so seven service packs isn't unreasonable. Especially as the service packs had extra functionality in them as well.
Windows 2000 & XP have moved towards individual updates rather than making you wait for service packs. Instead of hunting for drivers etc you can now get your machine to do all the work for you ( if you want ).
> the fact that OS3.9 only had 2 is pretty impressive in my mind -
Ah, I get it now. So it's ok for OS3.9 to have two updates, but not Windows 2000.
And here's your starter for ten points:
Why did the original Amiga 1000 require you to load Kickstart from disc.
Was it:
A: They had ordered too many floppy disks and needed a way to get rid of them.
B: The bootup jingle, followed by the disk crunching was found to turn women on.
C: The OS was so full of bugs they were sending out updated disks every day.
OS1.x didn't stabilise until 1.3, even then there were two or three disk updates. The same thing happened with OS2.0, that didn't stabilise until 2.04.
Commodore went under before they could release any more updates for OS3.x. Which is why we're stuck with Kickstart 3.1, not because it's bug free.
All software has bugs, bug fixes aren't the enemy. Windows has more functionality, therefore it will have more bugs. Now you can hate Microsoft for whatever reason you want ( I don't like their business practices very much ), but as a software developer I don't see that they are any worse than the rest of the industry. In some cases they are better than others, at least they fix their bugs.
Now you can flame me if you want because I use a Microsoft OS, but I've only had my laptop since October when I had to give in as the a1200 just didn't cut it anymore. If you want a hassle free life then you're pretty much limited to Windows XP on an x86 Laptop. The only problem I have now is that my laptop is "obsolete". The mistake Commodore made is they didn't obsolete their hardware quick enough.
Phill
New Hans-Joerg Frieden interview : Comment 23 of 69ANN.lu
Posted by Phill on 15-May-2002 09:26 GMT
In reply to Comment 16 (Akaru):
> Geez if Microsoft actually lived by those rules they'd have a decent product,
> as it is they're selling some rank ass dog food that is 90% chalk, and 10%
> horse uhhh....
If you look at the products the developers are going to be using ( Windows XP & Developer Studio ), they are generally regarded by most 3rd party developers to be amongst the most stable & easy to use development environment.
Love them or loathe them, they do make decent development tools.
Phill
New Hans-Joerg Frieden interview : Comment 24 of 69ANN.lu
Posted by smithy on 15-May-2002 09:50 GMT
In reply to Comment 14 ([JC]):
>With LVO's, you have to assign them an offset, which effectively allocates
>space in a jump table. With symbol based linkage, you mark the symbol as
>exported, and that's it. It's easy, it's fast, and what's more there's no
>looking up what LVO does what.
I'm no expert on ELF, however, doesn't the ELF shared-library format use a symbol table, effectively a jumptable? The only difference is that in the current AmigaOS model, the offsets to the correct position in the jumptable are hardcoded at compile-time. Whereas with the ELF model, the method names (the symbols) are associated with a pointer giving the freedom for the library to move its functions around, or insert new ones at the beginning. The ELF model is a touch slower than AmigaOS, since there is a need to do lookups.
Anyway, the point is, that unless you're coding in assembly you'd not know the difference, and if you were, the symbol table would be slightly harder to use!
>Furthermore, it would be in Amiga's interest to follow a common EXE format
>like ELF. This increases the chance of people developing tools for it, and
>also means existing tools can be ported relatively easily.
I don't see that there is any reason to use ELF, or adopt what are essentially any other UNIX standards. Microsoft's COFF is more widely used than ELF, should we use that instead? I often boggle at the mad rush to adopt UNIX standards. We should remember that UNIX people proclaim them to be industry standards and they are interoperable - but in fact they are only standards and are interoperable amongst UNIX platforms.
New Hans-Joerg Frieden interview : Comment 25 of 69ANN.lu
Posted by Leif on 15-May-2002 10:22 GMT
In reply to Comment 24 (smithy):
Havent it been gone threw many times already, that the new
librarysystem will actually be faster than before ??
As I understand, the look-up of symbols are done *loadtime*
(loadtime linking), making runtime calls as fast as any
other function residing inside the exe.
Leif
New Hans-Joerg Frieden interview : Comment 26 of 69ANN.lu
Posted by Anonymous on 15-May-2002 10:37 GMT
In reply to Comment 25 (Leif):
> As I understand, the look-up of symbols are done *loadtime*
> (loadtime linking), making runtime calls as fast as any
> other function residing inside the exe.
If that is the case, I have three questions:
a) Does this mean the exe format used has information that allows the loader to link to any and all system libraries on load (how is that done?)
b) In the current AmigaOS you can choose not to open a library until you need it. If the link occurs on loading, this cannot be done?
c) Does this mean we have the idiotic system I understand is used by Linux where newer versions of a library are incompatible with older ones unless the program is recompiled to use the newer one?
New Hans-Joerg Frieden interview : Comment 27 of 69ANN.lu
Posted by Björn Hagström on 15-May-2002 11:01 GMT
In reply to Comment 26 (Anonymous):
"c) Does this mean we have the idiotic system I understand is used by Linux where newer versions of a library are incompatible with older ones unless the program is recompiled to use the newer one?"
Not as long as no changes are made to the functions used (That is, the parameters and functionality). In other words, it would be as possible to get incompabilities in this new system as it is in the old jump table system. If the developer decides to change the API of old functions then you would have an incompatability on your hands. There is no difference in this respect between the two different methods of library calling.
You raised some interresting questions in a) and b) (they are basically the same). And I can only complement them by asking the Hyperion guys:
Must all libraries used by an exe be loaded into memory before the onload linker can do its job? Or can linking/name lookup be done in real time as well?
/Björn
New Hans-Joerg Frieden interview : Comment 28 of 69ANN.lu
Posted by Leif on 15-May-2002 11:07 GMT
In reply to Comment 26 (Anonymous):
a) Does this mean the exe format used has information that allows the loader to link to any and all system libraries on load (how is that done?)
QNX does this..
b) In the current AmigaOS you can choose not to open a library until you need it. If the link occurs on loading, this cannot be done?
You could probably still Open/Close it..
hmm.. very good point, there could be some cases things
get stange.. or pherhaps not.. depends on how its implemented.
c) Does this mean we have the idiotic system I understand is used by Linux where newer versions of a library are incompatible with older ones unless the program is recompiled to use the newer one?
Cannot see why it should mean that.
New Hans-Joerg Frieden interview : Comment 29 of 69ANN.lu
Posted by Leif on 15-May-2002 11:24 GMT
In reply to Comment 28 (Leif):
Okey.. Those who worked on theese thongs in AOS4
have said that symbols are used instead of a jumptable.
They have said that one (1) instruction is saved
everytime a function is called. This instruction
I then guess must be the "jmp xxx" located in the
jumptable for the old libraries. So, in the new libraries
we call the functions directly. How is that possible,
we cannot know its adress (its in a different binary) ?
SOme kind of linkage is done, I would strongly belive
its done *loadtime* (runtime could be a bitch :)).
So, somewhere in the exe we must provide information
about at which places in the code we call lib-functions
and which functions we call. Similar to relocation,
the OS puts the addresses of the lib-functions at
the right places in the exe. But there are other things
like open/close, will it be needed or not, probably not,
but it could be handy the library itself to get an openLibrary
call from the exe so that it may set up things for it.
Im clueless in this area as to how it will be done.
Hopefully better anyway :)
New Hans-Joerg Frieden interview : Comment 30 of 69ANN.lu
Posted by Gabriele Greco on 15-May-2002 11:34 GMT
In reply to Comment 28 (Leif):
With ELF you can do a lot of nice things, expecially if you link with -rdynamic. In that mode you build also a symbol table for your application and you can for instance replace a shared library function call providing one with the same name in your application or provide it to make your software compatible also with older version of a certain library. The function bindings are done at load time, so it's faster than the actual amiga format. You have to bind the call only once, then is like a simple function pointer. I'm not that used to know assembly 680x0 or ppc clocks, but I'm quite sure that something like:
move FunctionAddress, a0
jsr (a0)
Is faster than:
move XXXBase, a6
jsr LVOFunction(a6) -> jmp #FunctionAddress
...the first method is the one you use at runtime both with ELF and, AFAIK, OS4.
Sorry for mispelled asm but I mostly code C :)
New Hans-Joerg Frieden interview : Comment 31 of 69ANN.lu
Posted by Leif on 15-May-2002 11:43 GMT
In reply to Comment 30 (Gabriele Greco):
The problem with above call is that no environment for the
library itself is provided (librarybase).
Ofcource, some functions dont need it, but many do.
Maybe that thing is takken care of other ways though
( the lib uses pc-relative data.. but then we have no
way of multiple instances of global library-data)
I guess its proc AND cons with the new concept.
Saving ONE instruction is not very much to throw
away other advantages though.
New Hans-Joerg Frieden interview : Comment 32 of 69ANN.lu
Posted by Kronos on 15-May-2002 11:45 GMT
In reply to Comment 29 (Leif):
If the library are loaded (and linked) by LoadSeg()
I see (atleast) one possible problem.
If your SW uses optinal (third party) libraries and
will simply turn off some functionally when this
library isn't found, how is the loader going to know
what is needed and what is only optinal ?
New Hans-Joerg Frieden interview : Comment 33 of 69ANN.lu
Posted by Daniel Allsopp on 15-May-2002 11:45 GMT
In reply to Comment 23 (Phill):
>If you look at the products the developers are going to be using ( Windows XP >& Developer Studio ), they are generally regarded by most 3rd party developers >to be amongst the most stable & easy to use development environment.
I currently use at home and at work, Windows 2000 Professional SP2 and Visual Studio 6 Enterprise Edition SP 4 and together they make the most stable and enjoyable development platform.
I've recently switched my game development to my PC using the above software combination and it's so much easier and nicer to use, compiles in <20 secs, unlike my GCC StormC which takes approx 5-10 minutes, even on my 060!
>Love them or loathe them, they do make decent development tools.
They do!! They might be a bunch of monopolistic tossers but they make nice software, i.e. VisualStudio, Office 2000, Money etc etc...
New Hans-Joerg Frieden interview : Comment 34 of 69ANN.lu
Posted by Leif on 15-May-2002 11:50 GMT
In reply to Comment 32 (Kronos):
Good question :)
Suppose the loader then just doesnt link in that library..
This would be okey, if there is an Open/Close scheme,
as we then couldnt open the library, just ilke the old way.
New Hans-Joerg Frieden interview : Comment 35 of 69ANN.lu
Posted by Kronos on 15-May-2002 12:08 GMT
In reply to Comment 34 (Leif):
The problem is would make the whole thing
rather complicated as you can't fill in
the symbolic-links after the SW is started
You would need to go back to the old jump-
table-system.
New functions won't be included to it.
Third-party libs would made "classic" or
"SG" and probraly not both.
If they are really doing it this way (doubt it)
!!!Why fix it if it ain't broken ?!!!
New Hans-Joerg Frieden interview : Comment 36 of 69ANN.lu
Posted by Leif on 15-May-2002 12:12 GMT
In reply to Comment 35 (Kronos):
It ceirtanely not isnt broken :)
If they use ELF, maybe they want to use its features.
Well, they should have a good reason, otherwise why bother.
New Hans-Joerg Frieden interview : Comment 37 of 69ANN.lu
Posted by Kronos on 15-May-2002 12:16 GMT
In reply to Comment 36 (Leif):
AFAIR T.Frieden wrote on the AOne (or was OS4 ?)
mail-list that they would use an enhanced (Amiga)-
hunk format and not ELF (like PuP/MorphOS).
New Hans-Joerg Frieden interview : Comment 38 of 69ANN.lu
Posted by Gabriele Greco on 15-May-2002 13:03 GMT
In reply to Comment 37 (Kronos):
Anyway I don't see why many people see unixlike or windowslike solutions BAD without even knowing how them work....
I agree that often the windows API are really ugly and unconsistent between subsystems, but many unix solutions (ELF, pthreads, device usage) are really nice, easy to use and powerful together, that's a rather difficult task. So I really don't understand why ELF should be bad only because it's an "alien" format. ELF is newer and more flexible than the OLD amiga hunk system, as pthreads or also windows threads are more configurable than Amiga task/processes...
I think that eventually Amiga should do something similar to what apple did with OSX: an AmigaOS layer on the top of a solid unix layer...
BTW: Amithlon is quite near to this solution, the only problem is that there is no native OS for x86 (nor OS4, nor OS3, nor Mos) and that to write little endian apps on a big endian machine it's really messy and not very clean indipendently on how good you can hide this with the compiler.
New Hans-Joerg Frieden interview : Comment 39 of 69ANN.lu
Posted by Kronos on 15-May-2002 13:19 GMT
In reply to Comment 38 (Gabriele Greco):
>MacOSX
Have a look at MorphOS (AFAIK Quark is somewhat
similar to UNIX).
There is a (cross)compiler for Amithlon that does
use the "right" endian-format and it is getting
better allmost every day.
New Hans-Joerg Frieden interview : Comment 40 of 69ANN.lu
Posted by Ole-Egil Hvitmyren on 15-May-2002 13:59 GMT
In reply to Comment 38 (Gabriele Greco):
I disagree. I say it's better to think new instead of just copying everyone else. If OS5 happens, let it not be based on on any unix, but written from scratch to accomodate excactly the things Amiga wants, not what worked so-so for Apple. However, I don't claim to have the right opinion, just to have the right to an opinion.
New Hans-Joerg Frieden interview : Comment 41 of 69ANN.lu
Posted by Just Say NO on 15-May-2002 14:10 GMT
In reply to Comment 22 (Phill):
Hey, you could've purchased a TiBook and ran OS X. Running
XP was simply a total capitulation to M$. I don't buy your
reasoning on that one. Fortunently there are still some
people with enough backbone to "JUST SAY NO" to M$.
New Hans-Joerg Frieden interview : Comment 42 of 69ANN.lu
Posted by Ole-Egil Hvitmyren on 15-May-2002 14:19 GMT
In reply to Comment 41 (Just Say NO):
The problem with a lot of the people who say no to Microsoft is that they get caught (how the heck are you supposed to spell that anyway?) up in this "non-free and or non-open-source software must be eliminated"-crusade, effectively ruining life for anyone who could have been a competitor to Microsoft. Linux will forever be a bedroom-coder OS, because it is made by bedroom coders FOR bedroom coders. I 'm not throwing shit at Linux, just saying some of it's followers should be a bit more open to people who are trying to make a living off of software. Opera, Acrobat Reader and Eagle (electronics-suite) immediatly springs to mind. It takes manpower to come up with brilliant software, even if some coders can cook up a good "alternative if you don't mind the odd crash" in an evening. But please, do say no to Microsoft. Just don't become fanatical :)
New Hans-Joerg Frieden interview : Comment 43 of 69ANN.lu
Posted by reflect on 15-May-2002 15:14 GMT
In reply to Comment 21 (Robert "RAG" Gustavsson):
Actually, hotmail migrated to win2k after some 3 years of trying to move it away from fbsd/solaris.
They finally moved it to w2k and opened up holes big enough to fit tankers in them..
New Hans-Joerg Frieden interview : Comment 44 of 69ANN.lu
Posted by smithy on 15-May-2002 15:17 GMT
In reply to Comment 38 (Gabriele Greco):
>I agree that often the windows API are really ugly and unconsistent between
>subsystems, but many unix solutions (ELF, pthreads, device usage) are really
>nice, easy to use and powerful together, that's a rather difficult task.
>So I really don't understand why ELF should be bad only because it's an
>"alien" format. ELF is newer and more flexible than the OLD amiga hunk system,
>as pthreads or also windows threads are more configurable than Amiga
>task/processes...
Most UNIX technologies are file-based, and don't integrate well with AmigaOS's
device-based APIs. Just look at how badly the BSD Socket library integrates
with AmigaOS. Trying to merge two completely different systems of event
handling created headaches and inconsitencies.
UNIX technologies may be very nice, but without turning AmigaOS into a UNIX
clone, they cannot be used easily, or nicely.
New Hans-Joerg Frieden interview : Comment 45 of 69ANN.lu
Posted by Nicolas Sallin on 15-May-2002 16:39 GMT
Kronos,
Quark has nothing to do with Unix. It's a custom kernel, with some
ideas coming from L4. But its API and design is custom anyway.
Otherwise, AmigaOS4 uses ELF and they explained on comp.sys.amiga.game
how they extended exec api with some functions to handle&research the
symbols.
ppc.library already does all that since 4-5years btw :-)
New Hans-Joerg Frieden interview : Comment 46 of 69ANN.lu
Posted by Kronos on 15-May-2002 16:48 GMT
In reply to Comment 45 (Nicolas Sallin):
O.K. ;)
But I'm quite Thomas did write about an
enhanced version of the Amiga-hunk-format
on one of those maillist shortly before
he left the public.
New Hans-Joerg Frieden interview : Comment 47 of 69ANN.lu
Posted by Leif on 15-May-2002 17:02 GMT
In reply to Comment 46 (Kronos):
Are you thinking about WarpUP pherhaps ? :)
It uses EHF, extended hunk format.
New Hans-Joerg Frieden interview : Comment 48 of 69ANN.lu
Posted by Kronos on 15-May-2002 17:05 GMT
In reply to Comment 47 (Leif):
Nope he was talking about COS4.
New Hans-Joerg Frieden interview : Comment 49 of 69ANN.lu
Posted by Kamel Biskri on 15-May-2002 19:44 GMT
In reply to Comment 45 (Nicolas Sallin):
Hi There,
Quark a kernel ? You probably mean Darwin no ?
Kml
New Hans-Joerg Frieden interview : Comment 50 of 69ANN.lu
Posted by Alkis Tsapanidis on 15-May-2002 19:47 GMT
In reply to Comment 49 (Kamel Biskri):
Eh, Quark is the MorphOS kernel:)
Anonymous, there are 69 items in your selection [1 - 50] [51 - 69]
Back to Top