If you want a more in depth overview after looking at the httpget.c example:

The "bible" for socket style network programming is W. Richard Stevens: UNIX Network Programming. See http://www.kohala.com/start/unpv12e.html

Most of it translates effortlessly to Amiga bsdsocket code.

@gregthecanuck

You need either a #! or similar header OR the script/execute bits, or there'll be no safe way for the system to separate a script from just any text file (datatypes can't help you there). Relying on #! of course would break practically every AmigaOS script in existence, but requiring either #! or script/execute ought to be safe enough.

I'm not sure where I'd stand on not requiring it to be set executable - but I agree a separate script bit isn't really necessary.

@trixie
Quote:
[/quote]
I haven't ready this entire thread yet, but I just wanted to mention this:

After starting to use AmigaOS again after 12 years away, during which I've used mostly Linux, some OS X and a tiny bit of Windows, the ASL requester is one of the things I've enjoyed the most...

Compared to the other file requesters I've had to endure, it's by far the most comfortable to use. The main thing I miss is a preview option for images.

It may be ill-designed, but most file requests in general are horrendous.

@ChrisH

Quote:

It's a little bit different, though, in that if you want to keep using your classic machines without a PPC card you're kinda limited.

And for those of us stuck on x86 boxes for now, the alternative is 3.x under UAE or AROS. Personally I use both. Though the stuff I'm playing with under 3.x is stuff I hope to make portable to AROS and OS4 too.

@ChrisH

You could use it to pretty much take total control of the classic. Zorro is pretty much an extension of the system bus. The side extender port on A1000 and A500 as well.

The XMOS should be fast enough to be able to monitor bus access in realtime and trigger on specific events. You could for example, if you wanted to, halt the CPU on certain memory accesses, or keep a snapshot of the values written to the custom chips (which is how cartridges on the Amiga were able to do accurate freezes/snapshots of games etc., despite the number of write-only registers), or DMA stuff straight in/out of memory on the classic, or emulate an expansion card to let the classic use the X1000 as a hard disk or memory expansion (wouldn't be hugely fast, but then again the maximum speed imposed by the typical transfer rates on the side expansion / Zorro isn't exactly high either.

Of course the real answer as to why one might do something like that is because it's fun and can be done :)

@nicholastaylor

The first challenge is going to be the POSIX/pthreads bits in the JIT itself. ixemul + the AOS4 pthreads implementation might get you reasonably far.

The biggest challenge, however, would be porting the various native parts of the runtime libraries. GUI toolkits etc.

It depends a lot what your goals are. If you "just" want Mono on X1000, that's a *lot* easier than getting something complete enough to run most C# apps. The former could be done with thin mappings to AmigaOS, while the latter requires supporting a lot of other libraries as well.

@Helgis

Quote:

I'm not talking about Xorro. I'm talking about the Zorro bus on the classic Amiga's (including the side expansion port on the A500 / A1000). This is more or less a buffered extension of the M68k system bus, complete with support for bus mastering DMA.

As such, you can almost entirely control the machine if you connect a fast enough device (like the Xena / XMOS) to it.

What I was describing was briding Xorro to Zorro, to use the IO pins on Xena to effectively control the system bus on the classic.

@QuikSanz

I *think* that it'd be almost trivially for real Amiga's, since they expose the Zorro bus either internally or as a side expansion. Only issue might be number of IO pins required - might have to multiplex access. If not, it ought to be more or less a connector on either end and a cable, possibly some voltage regulation.. Zorro is more or less just the plain system bus (that's why Action Replay can freeze an app and recover the settings of special registers, for example - it just listens to the bus and "mirrors" the writes done to the various custom chip registers).

You could "slave" the classic off the X1000 completely quite easily, I think, including remotely rebooting it, halting the CPU so you can manipulate memory (load/save, whatever) and custom chip registers, or simulating various devices, including bridging hard disk access etc.

@Spirantho

Quote:

You're sort of wrong. It's first and foremost a programmable I/O chip.. It has support that makes it easy to run in massively parallel settings - for example they have a board where 16 quad core chips are connected in a 4D hypercube so all the cores can send messages to eachother. That part is very much like the transputer.

But the cores are not nearly powerful enough to compete in terms of raw processing power with "normal" server / desktop CPU's - it'd take a whole stack of them to even approach the raw processing power of the PowerPC in the X1000, for example (regardless which CPU they're planning on using).

What they're good for is stuff that requires programmable IO pins and/or very low latency (they can handle dozens of IO events before the PPC could even handle a single interrupt). E.g. if you'd like to hook up old Amiga hardware it would be a matter of a software driver, and a practically empty PCB just ensuring voltage regulation etc. and providing a suitable connector.

(just got the crazy idea of bridging an expansion port from a classic Amiga to the X1000, since you can easily control the bus fully including halting the 68k, monitor read/writes "live" etc... Not sure what I'd *use* it for, but it'd be fun - I do think it'd easily allow DMA'ing into the classic's memory too, for example - Zorro is "easy" to interface with..)