Hello John-Mark On 07.01.11, you wrote:
> On Thu, 2011-01-06 at 17:49 +0200, Bernd Roesch wrote: > >> netsurf MUI GUI was do from itix.But it does now not work, because of >> lots changes in netsurf. > > It's in a branch in SVN, if someone wants to resurrect it. I doubt it's > that much work to do so -- the interface between NetSurf's core and > frontends has not changed a very great deal (and, where it has changed, > it's for the better). I'd imagine it'd be about a week's work to fix it > up, should someone want to. we first begin to port MOS netsurf with MUI gui to amiga os3.the compile is ok, only crash happen.we need then replace MUI4 function with MUI3. But we do nothing more then, because it get public that plotter interface change. Itix too have no motivation to change his MUI Version to new plotter interface, so we use after some time the SDL Version. Artur begin then to add Video Features download etc to netsurf. only miss for a good browser in SDL Version is java script and frames. netsurf use MUI 4 GUI but because OWB and netsurf GUI is in part the same, and OS4 MUI API is same as OS3 MUI API , and OWB MUI from MOS is port to OS4, this maybe help also netsurf MUI GUI. Artur try to get OWB OS4 Port working now.this is based on a webkit core with java and frames. then we see if and how much netsurf is faster. sorry to say that, if netsurf is not faster, i think we have no motivaton to spend work in netsurf when it have no java script and frames. here is AROS download and you can see that it is availible for many platforms.so when netsurf on MUI work, it can work on all this platforms and of course on OS4 too. http://aros.sourceforge.net/download.php#nightly-builds But Chris want his reaction and this is only here for 68k there is class act, it is free but i dont know if netsurf OS4 can work with this. http://aminet.net/package/dev/gui/classact33 > >> binutil is old, here amiga target is remove in mainsource, but because >> its only a assembler and linker this doesnt matter. all GCC can work >> with that. > > Aside from the fixes to mainline binutils I had to backport to get it to > assemble the output from modern GCC, of course. wy does it not compile ? maybe you use the wrong compiler.I have not test if the GCC 4.5.1 from AROS 68k.maybe it output elf. We compile SDL netsurf with cygwin GCC 4.5.0 and binutil 2.14. i do the GCC ports to amiga os 68k, and i have also instructions include how build newer GCC(or GCC for a special linux) and new libs as openssl 1.0.0 (Artur use for netsurf) http://amiga.sourceforge.net/ I can also send you full netsurf SDL dir with all libs, so you can see that binutil is ok for netsurf. In fact we know no program that fail to compile because of binutil when i have time i plan add a elf loader(which is in AROS 68k too)and use the 68k elf binutil. because GCC 4.5 link time optimization work only on elf and the 68k amiga use the removed Linux format (forget name) before Linux change to elf. > >> clib2 does not work good, it miss lots functions. > > It needed stub implementations of 7 functions to get NetSurf and all of > its dependencies (including the stage 2 toolchain) to compile. Aside > from vfork(), they are all obscure BSDisms that aren't all that > portable, anyway. On that basis, it's fine for our needs. the problem is C++ Programs, netsurf (and libxml) do lots and freqeuently do small mem alloc and free.amiga OS 68k have only a first fit allocator.this get lots fragmentation and heavy slowdown after longer netsurf use. there is also a tlsf mem allocator here, but its always slower because good fit do lots of cache misses. buddy allocator in ixemul is fast and give no fragmentation slowdown on netsurf ixemul can work with two memallocator, a internal buddy allocator and default amiga allocator. so can see that when ixemul not use the buddy allocator, netsurf run slower and slower after longer use. its extrem see here.10 * this page reload slow down by 4* http://www.fotocommunity.com/search?q=&index=users > > > J. > > > > Regards
