Building Netsurf 3.x for 68K

[wawrzon]

Indeed. Everyone in the Amiga community keeps yelling that we need this, we need that, but almost no one does anything (including myself ).

sad is, when people who can do something, are yelling instead of doing it actually. i for my part at least try to contribute, even if my abilities are minimal. imho, its hard to maintain motivation and pull anything all alone, better to join together in open projects.

That may have several reasons, and we'll probably never know.

most likely its simply a lot of work.

Which will just lead to bad performance on 68k. We've already seen that.

features come at cost, so no wonder its slower than a browser that imply doesnt parse css. i dont know, but i imagine for an experienced coder it may be easier to tweak an open engine instead to write it from scratch. im not talking about current webkit, due to its high resource demend, but netsurf could actually be an option. even if it is also not perfectly keeping up with web standards. (you can check it on a pc as well).

[wawrzon]

Sadly not, because it's 3.5+.

yes, its pita that chris insists to check for libs and reaction classes versions manually, for me it might work on aros68k if not that. but its understandable as he doesnt want a trouble with supporting the genuine system. he codes nostly for os4.

[Thorham]

yes, its pita that chris insists to check for libs and reaction classes versions manually

You have to. If you're using Reaction specific features that don't exist in 3.0/3.1 then it'll simply crash. You always have to request libraries of the right minimum version when writing software.

It's different, for example, when you're using the AmiTcp4 SDK while you're only using features that AmiTcp3 supports. Your software won't run with AmiTcp3 unless you recompile it with the AmiTcp3 SDK. NtpSync is an example of this. Works perfectly fine with AmiTcp3 when you recompile it (luckily it includes the source code).

However, in the case of Netsurf, it probably simply can't work on 3.0/3.1. Wouldn't be entirely surprised if it can't work with Aros either.

[utri007]

iBrowse doesn't actually render amiga.org in 20 seconds. It starts to display web page when some pictures a loaded, it still loads pictures quite much longer, but web page is useable.

I think that Chris doesn't have anything against it that somebody would fix Netsurf to use class act. He just don't want to do it.

@Thorham I'm little curious, why not OS3.9? What kind of Amiga you have? I do know that it is possible to make OS3.0/1 perfectly useable, but it It just easier with OS3.5/9.

[Thorham]

iBrowse doesn't actually render amiga.org in 20 seconds.

On my A1200 with Blizzard 1230 MK4 64mb it most certainly does. After 20 seconds, it's completely done (32 color double pal screen on an SVGA monitor).

I'm little curious, why not OS3.9?

Because it has nothing I want, and no software I want to use. 3.0 is fine. Kickstart 3.1 would be handy because my hard disk spins up a little to slowly, but that's it.

[wawrzon]

However, in the case of Netsurf, it probably simply can't work on 3.0/3.1. Wouldn't be entirely surprised if it can't work with Aros either.

it has not been tested. it is a development version, so it would make sense to disable the check and see what it actually needs. im not sure i have seen netsurf reaction frontend features that wouldnt work with the free available class act. but to be honest i could have compiled netsurf commenting out the checks myself, except its low priority, because netsurf build system needs so much fiddling with before you can produce an executable and start to edit the code.

[utri007]

But to be honest i could have compiled netsurf commenting out the checks myself, except its low priority, because netsurf build system needs so much fiddling with before you can produce an executable and start to edit the code.

https://github.com/DNADNL/NetScript

You can esily compile it and play with code.

[wawrzon]

https://github.com/DNADNL/NetScript

You can esily compile it and play with code.

"easily".. i love these comments.. have you even tried that? then why none uses it? this script is for cygwin, and im not installing cygwin again, if you know what i mean. i even ran it and looked it through on my vmware lubuntu and since it made some tweaking necessary postponed it.

[utri007]

"easily".. i love these comments.. have you even tried that? then why none uses it? this script is for cygwin, and im not installing cygwin again, if you know what i mean. i even ran it and looked it through on my vmware lubuntu and since it made some tweaking necessary postponed it.

I use it and if I can use it, then it is easy to use.

[wawrzon]

I use it and if I can use it, then it is easy to use.

what are you running it on?

[utri007]

Windows 7

[wawrzon]

Windows 7

sigh... cant you anwer properly? so you are running it on cygwin. you simply followed the guide step for step, thats why its easy for you. what part of my comment: "this script is for cygwin, and im not installing cygwin again" was unreadable for you?

so, for the record, i tried that script again on my lubuntu build setup, and as i said it needs some tweaking. it doesnt build the gcc itself for some reason. i have gcc 6.2.0 amiga toolchain and i have gcc4.6.4 and gcc 6.3.0 aros68k toolchains, but i doubt it will build with these out of the box if the default is 3.x.x.
perhaps i will build cahirs compiler by hand later, and retry the script. but now i need to work. probably it ll just get postponed and forgotten another time, as there is more pressing issues elsewhere.

[utri007]

sigh... cant you anwer properly? so you are running it on cygwin. you simply followed the guide step for step, thats why its easy for you. what part of my comment: "this script is for cygwin, and im not installing cygwin again" was unreadable for you?

Usually guides are made to make things easier. So following the guide step by step is a default. I just told about my experience of it, I didn't ask anything from you, you asked.

First you try to be sarcastic, hinting that I haven't actually used it.

"easily".. i love these comments..

Then you ask your questions, hinting that it is so difficult to use and nobody hasn't actually used it.

have you even tried that? then why none uses it?

Of course I use cywin, only possible question was do I run it top of windows or linux.

As a conclusion about netsrcipt is, that it is easy to use if you follow included guide. It shouldn't be a surprice, if it is harder to get work if you don't follow guide.

[wawrzon]

I didn't ask anything from you,

here, as answer to my post:

https://github.com/DNADNL/NetScript

You can esily compile it and play with code.

you asked.

yes, i had to pull every bit of info from your nose, because i have tried that script before and didnt want to lose time with just setting the build system up again. it occures that i was right. i may look into that further, likely its trivial enough, but your general attitude doesnt really motivate me. anyway its off topic here.

[chris]

I think that Chris doesn't have anything against it that somebody would fix Netsurf to use class act. He just don't want to do it.

Correct.

it has not been tested. it is a development version, so it would make sense to disable the check and see what it actually needs.

I know what it needs as I have the source code and the Autodocs.  The versions requested are definitely correct for the libraries, for the classes maybe not - it looks like when I ported it back to 3.5 I've blanket set the version numbers to 44 (bevel.image, for starters, will be happy with any version).

 im not sure i have seen netsurf reaction frontend features that wouldnt work with the free available class act.

I'm not so sure.  I've definitely been using newer window.class features - although I reverted some of this or created workarounds for 3.5/9 - and chooser.gadget and clicktab.gadget are dodgy on OS3.5 (clicktab is broken also on OS3.9).  Even if it turns out I'm not explicitly using v44 features, it's highly likely some bug will mean they don't work as intended anyway.

I'll go through and set the correct version numbers when I have sufficient time to do so.

so, for the record, i tried that script again on my lubuntu build setup, and as i said it needs some tweaking. it doesnt build the gcc itself for some reason. i have gcc 6.2.0 amiga toolchain and i have gcc4.6.4 and gcc 6.3.0 aros68k toolchains, but i doubt it will build with these out of the box if the default is 3.x.x.

It should work with the aros68k toolchains, but you will most likely need to do some tweaking of buildsystem.git (and the NetSurf makefiles), as it will be expecting m68k-unknown-amigaos as the triplet, rather than whatever AROS is, so various things won't work.

If you build a "normal" Amiga m68k version of gcc it should work fine, no matter what version, however you might need to re-point some paths.  I didn't realise there was a working gcc 6.2.0 for Amiga m68k.  It would be good to get the NetSurf toolchain updated if you want to look at that?