Netsurf for 68k amigas, css capable web brower

[unusedunused]

One of the basic rules of benchmarking anything is to remove the uncertainty or your results will be totally useless. Everyone and their cousin knows this...

it count not the exact time, because in real world internet load time can differ around 1-2 sec on fast server(as the netsurf and other sites i name are.)

Everybody knows that when a page on a amiga/ winbrowser can load in 1-2 sec and OWB need for this 90sec(without java script), then the influence of internet acess is about 1-2% at overall time, or something is bad in code.

and only this value is usefull how much slower it is in compare to aweb or OWB and if it is usefull

>Have you actually figured out why it crashes? Surely developer of >your skill is able to do this?

i have no time and fun to do so, Artur build first libnix and ixemul version too that crash.I look on both, but soon i notice libnix crash earlier.

When i port a Unix program i choose always the best solution.and this is not use libnix and add the missing functions, this is use a bsd kernel on amiga os, (ixemul)

>After that, the very first move is to get rid of ixemul usage. It just >will not work reliably with AmigaOS code.

and wy not ?

you cant compare MOS ixemul with 68k ixemul.
MOS ixemul have 2 very bad bugs, that make it very unstable, i find and fix that.i can tell you what places make problems.

when you use pngcrush and a picture that ami_stuff send me, and the MOS largeread code is enable, then there is extrem slowdown (more than 10*).it happen only on 1 picture.i deactivate that(ixmeul V48 have this code not) and all work ok

#if LARGEREADS
   r = fp->_r;
   /* TODO: should limit to specific buffering modes? - Piru */
   if (resid >= r + fp->_bf._size) {

      /* empty current buffer first, if a

--------------------

then next is,

in __write.c last MOS source have disable  IXTTY_RAW.
line in V48 was this.

#define TTY_NLCR_ENABLE (IXTTY_RAW |  IXTTY_OPOST | IXTTY_ONLCR)

when do a fh = open(audio:xxxxxxx) and do write(fh,buf,len)

give crap audio data hear on ahi.

use the V48 code fix that.

also i get now information wy ixemul V48 cant work on MOS even if MOS is called that it can execute 68k software.

its due to emulation bug.yes i call it a Bug, because when the AOS supervisor call is not support by 68k then MOS should nicely stop the program and tell supervisor is not support on 68k.

here is log output i get from a user and symbolinfo is in ixemul in.Here you can verify yourself hits come from supervsior if you dont believe that.

-> mlnet Hunk 1 Offset 0x000d0fe8

PPCStackFrame History:
StackFrame[ 0].LR-> Address 0x25660c70 -> mlnet Hunk 1 Offset 0x000c46d0
StackFrame[ 1].LR-> Address 0x25661860 -> mlnet Hunk 1 Offset 0x000c52c0
StackFrame[ 2].LR-> Address 0x2566d5bc -> mlnet Hunk 1 Offset 0x000d101c
StackFrame[ 3].LR-> Address 0x25668efc -> mlnet Hunk 1 Offset 0x000cc95c
StackFrame[ 4].LR-> Address 0x25668f50 -> mlnet Hunk 1 Offset 0x000cc9b0
StackFrame[ 5].LR-> Address 0x2565c2dc -> mlnet Hunk 1 Offset 0x000bfd3c
StackFrame[ 6].LR-> Address 0x2565c3e0 -> mlnet Hunk 1 Offset 0x000bfe40
StackFrame[ 7].LR-> Address 0x22f0bd58 -> MOSSYS:LIBS/ixemul.library Hunk 1 Offset 0x000088f0
StackFrame[ 8].LR-> Address 0x22f156b8 -> MOSSYS:LIBS/ixemul.library Hunk 1 Offset 0x00012250
StackFrame[ 9].LR-> Address 0x22f0bc1c -> MOSSYS:LIBS/ixemul.library Hunk 1 Offset 0x000087b4
StackFrame[10].LR-> Address 0x10121970 -> Module Hunk 0 Offset 0x00021970
StackFrame[11].LR-> Address 0xdeadfee1 **Not Valid**


MOS ixemul use that code and i guess its a bad hack and can break compatibility, when just set sr to 0 instead of getting the correct value.

#ifdef NATIVE_MORPHOS
  sr = 0;
#else
/*#ifdef MORPHOS
  if (has_morphos)
    sr = 0;
  else
#endif*/
    {
      asm volatile (" \n\
   movel a5,a0 \n\
   lea   Lget_sr,a5 \n\
   movel 4:w,a6 \n\
   jsr   a6@(-0x1e) \n\
   movel a1,%0 \n\
   bra   Lskip \n\
    Lget_sr: \n\
   movew sp@,a1        | get sr register from the calling function \n\
   rte \n\
    Lskip: \n\
   movel a0,a5 \n\
       " : "=g" (sr) : : "a0", "a1", "a6");
    }
#endif


I change ixemul now on two places to use AOS command.setsr.

important is that use gcc inline LPT stubs to not change flags on call
 
  sr=SetSR(0,0);

and some tests show this place make no problems.ixemul V61.2 contain now no supervisor calls.

now problem when run under MOS is ixpipe.the MOS version have add code and fake a wbstartupmessage, because MOS ixemul want read on ixpipe-handler wbstartupmessage.

fake a message i think is a unclean solution and i dont want 2 diffrent ixpipes to test more easy with V48 and V50 and above.

/* pr_CLI is NULL since we are a handler, so ix_open will expect a wb message. Fake one. */
  port.mp_Flags = PA_IGNORE;
  NEWLIST(&port.mp_MsgList);
  msg.sm_Message.mn_ReplyPort = &port;
  msg.sm_Process = &me->pr_MsgPort;
  msg.sm_NumArgs = 0;
  msg.sm_ToolWindow = NULL;
  msg.sm_ArgList = 0;
  PutMsg(&me->pr_MsgPort, &msg.sm_Message);

  ixbase = (APTR)OpenLibrary ("ixemul.library", NEEDED_IX_VERSION);

------

so new ixemul version need also correct ixpipe 68k for now.
more progress reports i dont get, but MOS OS4 is promotet to have a 68k emul and so it should execute 68k code and nothing else is ixemul.

[unusedunused]

Arent there working mui gui builders?
Wasn't there one called MUIBuilder?

yes but this are too old and not furtherdevelop and add too code in programs

i think it btw not usefull that a GUI is in program code.it should be loadable so when change something in GUI the program need not new compile.a large GUI compile need lot time.

on other systems there are gui editors and change GUI data need no recompile.

storm wizard offer this modern features too on amiga since long time, but sad, that it is not furtherdevelop

[Piru]

it count not the exact time, because in real world internet load time can differ around 1-2 sec on fast server(as the netsurf and other sites i name are.)

I guess you missed this before so I shall repeat it:

Solution: Remove the internet connection from the equation, use a locally cached copy of the page you want to benchmark. Only then you will be able to draw any kind of conclusions.

Have you actually figured out why it crashes? Surely developer of your skill is able to do this?

i have no time and fun to do so

Yet you have time to give bad advice and talk smack? How nice of you. You know it would be much more constructive to actually locate the bug and fix/workaround it rather than talk bull manure about ixemul and libnix.

When i port a Unix program i choose always the best solution.and this is not use libnix and add the missing functions, this is use a bsd kernel on amiga os, (ixemul)

Netsurf isn't a unix program anymore, at least if you want to have any kind of usable GUI with it (SDL one doesn't quite count as usable). ixemul is only usable for things that you don't actually port: for recompiles (configure && make). As soon as you start actually porting you get rid of ixemul as the first step.

After that, the very first move is to get rid of ixemul usage. It just will not work reliably with AmigaOS code.

and wy not ?

I've explained this before: AmigaOS code doesn't mix with ixemul. Regardless what you've supposedly done to "fix" this in your V600 ixemul versions does not really fix this fundamental design choice.

you cant compare MOS ixemul with 68k ixemul.

Yes I can. I've actually developed ixemul for years. I know the internals better than most. I can say for a fact that ixemul and AmigaOS code do not mix. Nothing I've seen from you has changed this fact, either.

--- Irrelevant drivel not worth commenting cut off ---

[unusedunused]

Yes I can. I've actually developed ixemul for years. I know the internals better than most. I can say for a fact that ixemul and AmigaOS code do not mix. Nothing I've seen from you has changed this fact, either.

yes you have develop ixemul for years, but what you have done is unclear.
The code in ixemul for MOS have code to build also a amigaos 68k versions.there are lots of ifdef morphos in.for example see in file uname.c

#if defined(__MORPHOS__)
static char sysname[] = "MorphOS";
#else
static char sysname[] = "AmigaOS";
#endif

but at end of all, the amiga os build from MOS source do not work with many programs on 68k.

So when you say you have done of ixemul MOS for years and not only do some few small fixes and test only with the few programs run on MOS or if one of the the few users report a Bug on MOS , i cant believe that you understand ixemul well, because the amiga os build run very bad, when do a simple compile. your current sourcetree with changelog is not visible for all, so cant see what you have done and what not.

>Yet you have time to give bad advice and talk smack? How nice of >you. You know it would be much more constructive to actually >locate the bug and fix/workaround it rather than talk bull manure >about ixemul and libnix.

yes i know that, but the MOS extension libnix source is not release for all as the old source, seem another weapon of red versus blue war.
I think MOS libnix source have this fix.

and i dont want have something to do with that stupid amiga PPC guy fanatic, that every side must reinvite the wheel again and no working together is possible.

this bad behaviour i dislike and i think was the main reason wy most of the millions amiga users are gone, and one good side of it, is that i never get the idea to buy such a PPC system, not that they use my money to build a cannon in the red versus blue war.I dont want support low quality system that want force users to buy the system with unclean methods.

and what some PPC OS devs say about ixemul is unclean.

So bring an argument wy it is impossible that ixemul run with amiga funcs ?

I like perfect systems and if its not perfect, i like when i can fix it to be perfect.

so if i need something fix allone i do that on the best solution that is here.and thats ixemul for 68k.

>I guess you missed this before so I shall repeat it:

as i told, eveybody can check with a fast windows PC and a fast windows browser(safari firefox) how fast it can be on his internet connect.and when he use amiga software and do the same test 5-10 times again then he get a average value and can say around 50% accurate how fast amiga browsers in compare to firefox etc are.

[Piru]

the amiga os build from MOS source do not work with many programs on 68k.

Now there's you problem: It was never intended to. You asked for the MorphOS ixemul source. This is what you got.

--- removed tons of irrevevant drivel, again ---

the MOS extension libnix source is not release

libnix is not released because it isn't open source.

--- removed tons of irrevevant drivel, again ---

eveybody can check with a fast windows PC and a fast windows browser(safari firefox) how fast it can be on his internet connect.and when he use amiga software and do the same test 5-10 times again then he get a average value and can say around 50% accurate how fast amiga browsers in compare to firefox etc are.

No they really don't. First of all the browsers run on a different system (different CPU, operating system, HDD and memory speed). Second the performance of the internet connection varies from time to time. Third the load of the remote web server varies from time to time. Regardless how many times you repeat the test you will not get any kind of comparable results that could be used to compare the browsers.

You can get better results by removing as much variables as you can. The obvious solution is to load the same web page from the local RAM disk. Even if you do this, you still won't be able to do any kind of comparison between different platforms (but you CAN compare the browser to another inside amigaos for example). To fix this you need to build Netsuft for the same system where the other browsers are tested (windows in this case).

I would have thought this is kind of obvious, but I you never know it seems.

[wawrzon]

@piru: we dont need to be that picky about the results atm. it might become important while the port will be optimized, but at the moment approximate results will do too. the most important is how netsurf does in praxis. and i have to say even with this simple sdl solution it is beyond my expectations. im testing the os4 post as well on the same system and have to say though more advanced os4 version seems a little slower on simple websites and less stable, which is but the fault of the system itself i suppose. os4 netsurf had stability issues from the start. on the other hand the current version renders correctly quite complex websites, which 68k port is not able to do yet.

[unusedunused]

on the other hand the current version renders correctly quite complex websites, which 68k port is not able to do yet.

can you tell me what pages OS4 netsurf render correct and 68k not ?

the netsurf build use the source from saterday, so it seem newer than MOS or OS4 build.

i see, in older netsurf use of new hubbub html parser was optional.

maybe on OS4 is used not hubbub, and when i report the site, netsurf hubbub can fix.

i dont believe that a wrong display website depend on sdl port or not.

the http://www.bahn.de is same wrong on OS4 ?

@Piru

>You can get better results by removing as much variables as you can.

yes of course everbody know that, but its good enough to get a quick overview about speed.for example on my winuae when i start netsurf 10* and for pageload measure  6 sec without JIT

now i start OWB several times and it take more than 90 sec without JIT.when i activate JIT, page is load always in 1,5 sec in netsurf

what do you think, is the inet connection a big error problem  ?.the rendering because of slow CPU have lots of % in overall performance.

and btw, i have some time ago test with a cache proxy (set env var in libcurl)and i notice with that proxy not faster page load.

and i think not that users now for test install a proxy.We can be happy to get any value.

you see from OS4/MOS side no value is come or what happen on reuters page, so i guess there is same problem.

>libnix is not released because it isn't open source.

Yes i see, there is no need to release the sources, because its no GPL.everybody can use it without releasing source.

how many % have MOS enhance in this libnix source ?
i think its a thing of fair play to make also availible what you get for free, when you use 20% of this code.since Genesi build no PPC Hardware or something other there is also no fight for bigger market possible.MOS is a retro system without new Hardware, wy do not share sources and have fun on amiga without much costs, and the fight for more users or bashing again features that the loved OS not have end  ?

I think you never test ixemul V61 on UAE and notice yourself, it run stable enough and far better than MOS native Version.

I for myself have test MOS, and i notice that MOS ixemul is not so compatible, and it stand also on MOS ixemul readme.

btw: MOS use too much AROS source, is since MOS2.x release some code give back to AROS ?.

here is libnix licence.

"""""
This package is public domain. That means that you can copy, use and modify it
without any problems and that you can get it for free. If you actually
paid for getting it this is completely your fault - I didn't see a cent
of that money. It also means that I cannot be made responsible for any damage
resulting out of the use of it - you simply shouldn't trust anything you didn't
pay for :-).
""""

[unusedunused]

Now there's you problem: It was never intended to. You asked for the MorphOS ixemul source. This is what you got.

Yes of course but when the MOS source contain so many ifdefs morphos and 68k asm code, so it seem it should run on amiga OS.

you never answer my question i ask severla times.

Have you really do much work and add this #ifdef morphos or have you only done some small fixes during last years ?

[wawrzon]

i dont believe that a wrong display website depend on sdl port or not.

the http://www.bahn.de is same wrong on OS4 ?

i have already reported it to you yesterday by mail. the problem looks the same. it is not a sdl issue.

[Piru]

libnix is not released because it isn't open source.

how many % have MOS enhance in this libnix source ?

Impossible to say. In overall it has been tweaked a LOT.

MOS use too much AROS source

Pardon? There's some kind of limit to as how much AROS source code you may use? The AROS source usage in MorphOS is limited to parts of dos, intuition, gadtools, diskfont, locale, commodities and some c commands. You can download the MorphOS 2.x ISO and check the documentation for the full list.

is since MOS2.x release some code give back to AROS ?

The license is obeyed, yes. MorphOS devs have been contributing back to AROS since 2001.

here is libnix licence.

"""""
This package is public domain.
""""

That is not the MorphOS libnix license. MorphOS libnix is not public domain. Some parts of the MorphOS libnix are based on the public domain work, however.

PS. Certain individual distributed MorphOS libnix sources with the wrong license at some point. This distribution was never authorized by the copyright holders.

---

Yes of course but when the MOS source contain so many ifdefs morphos and 68k asm code, so it seem it should run on amiga OS.

Assumption is the mother of all f***ups.

Have you really do much work?

Yes. I have. Now give it a rest already.

[unusedunused]

i have already reported it to you yesterday by mail. the problem looks the same. it is not a sdl issue.

yes i know and also you report that reuters page after some time do system crash on OS4 right ?

from my SDL Version tests netsurf dont crash system, but have the problems with some pages.i stress netsurf with reuter page showing go eating, after 20 minutes come back and it still want reload page.no crash see.


So i think ixemul work good enough.but its of course also intresting how performance is with libnix if it can run easy on it.

i see you have report the bahn problem chris in amigans Forum, so i need not report this on netsurf page.

so i guess this freeze problems come from the OS4 addons and its not netsurf fault.

intresting is now to know, if the reuter page on MOS do too freezes system.

[Fab]

now i start OWB several times and it take more than 90 sec without JIT.when i activate JIT, page is load always in 1,5 sec in netsurf

[...]

you see from OS4/MOS side no value is come or what happen on reuters page, so i guess there is same problem.

Despite all the arguments previously given regarding link speed/latency and whatever that can greatly ruin a "benchmark", here's a value for you:
OWB 1.3 MorphOS loads reuters in 10s on my peg2.
FireFox 3.something loads it in 15s on a pc/xp in the same network (and that pc is quite faster than my peg2).

So you see, it's easy to throw random values.

[lorddef]

I hope the handbags at dawn don't put you off this, and you keep going with it.

[unusedunused]

Despite all the arguments previously given regarding link speed/latency and whatever that can greatly ruin a "benchmark", here's a value for you:
OWB 1.3 MorphOS loads reuters in 10s on my peg2.
FireFox 3.something loads it in 15s on a pc/xp in the same network (and that pc is quite faster than my peg2).

So you see, it's easy to throw random values.

how do you measure that ?
My OWB 68k need 20 sec to show http://www.reuters.com page(800*600 window).but i want see how long it take until the complete page text is load so the buttom text can read (impressum).if there is a picture load later is not important, but whats important for usability of a browser is that the layout is correct as soon as possible and the text can read full early as possible.if then later the images are load take again 10 sec is not important.

owb 68k need for this until i can read  buttom text 85 sec

firefox need for this 4 sec.safari have same speed and ie too

[Fab]

how do you measure that ?

My measure is done when the page is fully loaded (images included). If I take your argument that not everything has to be loaded, then I can already scroll to bottom at 7s or so.

Anyway, silly test if it's not done locally, like Piru told.

[unusedunused]

My measure is done when the page is fully loaded (images included). If I take your argument that not everything has to be loaded, then I can already scroll to bottom at 7s or so.

Anyway, silly test if it's not done locally, like Piru told.

but when not all images need load to scroll to buttom, then not much data is need from internet and it is check if there are lockups or slowdowns and the browser is code good.

the amiga browser are in this case very clever programed, text is see very soon, no scroll block and images load later.

i still wonder wy there is no OWB on windows, there are patches since long time in OWB, only idea i have its far not so good as firefox or safari or other browser.

same is of course with netsurf too, but many pages are really same fast show as firefox and safari and IE on winuae.