FFmpeg + FFplay SVN-r18880 (m68k)

[bernd_afa]

I have absolutely no doubt that is what will happen - because ffmpeg specifically requires v61.

But consider a program that specifies v48. Compile it once with the v48 includes. Compile it again with the v61 includes. Run each of them under MorphOS. What will happen? If nothing bad happens, then I'll shut up and say no more on the matter. :-)

this work too.if a program crash because new features are miss, is not dependent from includes.

it depend on lib you link.the libc contain the functions for ixemul.

when you link with V48 libc then only the V48 functions can use.if a Unix program use a function that V48 not have, but V61 then you get a compiler error.the program cant build, no executable here, that can crash.

same happen when you use SDL for ixemul.on V48 libc.You get compiler error ix_CreateChildData not find.

when you use V61 libc stuff, then it compile well but on MOS get now error ixemul V61 need.

>I again refer back to the suggestion to integrate multiplatform >builds in the style of Jens Langner's projects.

but thats not my problem, we have ixemul on amiga sourceforge and everybody can jump in and do a MOS or OS4 native Port.

But there is red versus blue war and as can see on netsurf, OWB, or other ports that OS4 and MOS cant working together, same on ixemul lib too.MOS V49 and OS4 V51.1 are some kind of diffrent and diffrent sourcetree.

I ask some times about the OS4 source of ixemul, this i not get.

No, if you really hate MOS let those programs crash

i dont hate MOS.

[bernd_afa]

it was an ironic joke, meaning it should be better doing nothing (let it crash or whatever) rather than showing annoying requesters about a program [MAY] not work under MOS and making it die. (unless you create a crt0 which check for version AND revision of installed ixemul..)

on german anews there are more lies tell about ixemul V61 , but nobody tell wy this version not work.MOS os4 can run 68k programs.

So to have same effect as rename the library, a check for MOS and OS4 must done and the program end and if really a MOS or OS4 user copy the ixemul v61 lib to his system, then this is not load.

but of course nobody can test and find bugs wy this ixemul not work on MOS or OS4.

when enhance a amiga library for OS3, for example intuition library, or guigfx library  this make then too problems on MOS or OS4.

because MOS and OS4 have diffrent enhancement, that make only no problem because mos binaries cant execute on OS4 and OS4 benaries cant execute on MOS without emu.

but 68k executable can run on both.

So thats the reason that there some guys that want that 68k not furrtherdevelop.the degrade 68k to a retro system that should not enhance.And this is bad.

and this problem can either solved complete by rename the lib or by check for MOS or OS4.

how mos can detect i know, but how is OS4 detect in clean way ?

[bernd_afa]

Completely

We should not care what apache does in unix, instead if someone ports it over to Amiga, he should add the proper checks to malloc.

and thats the main reason, wy many devs left amiga, amiga devs should do more than other.they should make their program run on a 060/50 well or in case of MOS OS4 should run too on slow PPC 600-1 GHZ Systems.

When look what systems are currently on desktop, they are all more powerfull as this systems.so devs safe time, when they develop not for Amiga systems.

the Amiga OS does only very few grow in compare to other OS, there is red versus blue war.this mean user say sometimes new features are crap because their loved OS have them not(.This should demotivate the "enemy" side to enhance more to avoid trolling.so all in all AOS stay on low quality level.

Programs can not easy port from MOS to OS4 etc.some programs are done seperate by a OS4 or MOS OS developer or other, but very few working together.

Additional work cost time, and if a dev not love the amiga very much, wy he should spend additional time and stay on Amiga ?

On Unix there is virtual unlimited mem on amiga not.68k have luck so there is a virtual mem solution, but using virtual mem is slow.i dont like virtual mem, and notice suddenly much slowdown, so i better switch off.and many 68k users have no vmem active.

ffplay is over 12 megabyte sourcecode, this mean 12 million characters, now a porter should find out in this 12 million source lines which do the crash.

its hopless and much work, that Linux or windows progs work in low memory situation, because coders are so lazy here and dont check for enough mem, because of virtual mem system never can run out of mem.

I dont understand wy in the memalloc there is  a diffrence in compare to Harddrive, so i think this feature is same usefull as on hardrives standard.

when harddrive is full then a requester come and the user can free mem and continue, wy should not do same on memalloc by default ?

If some rare situations as apache without User dont need it, it can switch off.

without the feature  ffplay crash when show a digicam pic of size 3000*2000 when there are not 50 megabyte ram free.

i have more todo than fixing all Unix programs that do not check for enough mem.

and so nobody can report ffplay is crap because it crash showing a pic, this message was mainly add, because the other "sides" maybe dont like that 68k is furtherdevelop, and they look carefully in the hope to find something negative

So its no wonder that there are more and more left amiga and there is really less progress here and the disadvantage for amiga and other OS get more and more.

the reason that this happen is not the bancrupt of commodore i think , it is the split and the users who want fight that the commercial side get more users.

there are not left much here, so other side cant get much users that buy for much money a new amiga system.when i look on download for OS4 ffmpeg and ffmpeg GUI in OS4 depot, (newest version is only here and news links are to os4depot and its longer here)
it have 280 downloads.68k version come in the 6 days to 245 downloads and also some amikit downloads that contain this files.numbers i not know

so there is no reason that 68k should not furtherdevelop both systems are valid on more or less same dead or alive

amiga should make fun and not fight for all costs for users.
i dont want fight for system, i like use amiga dopus magellan2  and i think when i really spend much work on change ixemul name, there come other that bashing, so do only what go in fastest time.

[bernd_afa]

I find now a solution that is easy to add, and PPC User can happy that they have the higher library number ;-)

I open in crt0.o the ixemul library V48
when a lib is open, i check in library structure for V48 number and the revisionsnumber.

But the Question is, when the program run on MOS, the library is open, i must check if its V48.is there a correct AOS conform library structure here, or can i identify elsewhere that this is a ppc library and not the 68k ixemul ?

if i detect the 68k lib, then i check for revision

so newer versions with increased API get then a higher revisionsnum.

[bernd_afa]

Don't do this.

Yesterday, I started testing a method to open only the required library, meaning that if someone compiles a program using V61 SDK, and that program only uses V48 functions, the program will only require ixemul.library V48. that's nice/best, isn't it?
)

yes this work, when the libc and crt0.o from V48 is used.

but the arguments are from PPC side, if a program want install ixemul lib V61(from installer), MOS or OS4 show a requester if they should replace the library because existing is V49 and new is V61.

so i should not have a libnum over V49

of course thats not good argument, because there are no ixemul programs that install ixemul lib in lib folder with a installer.ixemul programs need no installer.i know only cubic ide with installer, but this install lib in a seperate folder on later search path.only is the ppc libnum is not good enough this lib is used.

also the correct issue is that MOS or OS4 print on install a warning, if the user want overwrite the library with a higher 68k Version, this can make problems.

win vista for example bring such a requester when you want install a XP driver.

But to say 68k libs should not have more features that need of course a libnum increase, or they should rename is really bad.