I hate Resolver

Does ROOL have access to the sources?

The blockage with all things ANT was most recently covered by Steve Revill in his London show 2018 talk. The path has become more twisted since ANT were bought out (after ROOL was formed) so it’s a lot of effort (on ROOL’s part) to keep restarting talks with another set of contacts who have no idea what they’re talking about!

All things considered do we really care about patching a multiply broken item?

https://www.riscosopen.org/bounty/polls/29 has no mention of Resolver!
Am I misunderstanding the Bounty or your comment?

There’s no mention, most likely because in TCP/IP bounty step 2 there’s no particular urgency to replace Resolver (or MbufManager or ShareFS). Resolver will need replacing by the time we get to step 4 (when IPv6 kicks in) though because it doesn’t do IPv6.

Premise: I don’t have the technical background necessary to code something of the complexity of an OS part.
But seeing all that discussion about Resolver, wouldn’t be simple to rewrite it from scratch? As a ‘black box’, the API (or something similar) must have been specified in some technical document.

That’s a sound premise, and exactly what I did for MimeMap.
All that’s needed is someone to step forward and actually do the rewrite, or for a bounty to be set up to collect to replace them (whichever happens first). In my view:

• MBufManager is the smallest/simplest of the ANT set, and quite practical to clean room reimplement
• Resolver is best put out to pasture as part of the TCP/IP bounties when needed, to avoid rewriting twice
• ShareFS is the big scary one, probably cheaper to hire a hotshot lawyer to get the existing code OK’d

you just tried to plug an unknown device into a network that you did not have permission to connect to and my RADIUS authentication said stuff off

Does ROOL not ‘own’ the ShareFS code then?

ROOL not ‘own’ the ShareFS code then?

No. The bits of code mentioned by Sprow belong to ANT, which is why an NDA is required before you can get access to the source.

Since the ‘Resolver’ module is in Assembly – we all have a copy of the code – but with no comments.

Is there a simple way of finding out what the calls are – to the BL’s to CLib?

*where address – gives somewhere in the ‘CLib’ module.
But which call!

Just out of interest, if Resolver makes calls to CLib, why do you think it is written in assembly?

https://espial.com/news/espial-completes-acquisition-ant-plc/

BTW, I love the enormous disclaimer that comes under the heading of “Forward Looking Information”.

‘Resolver’ module is in Assembly

It looks to me like a combination of Assembly and mostly C.

Can ROOL not purchase the suite for a nominal fee? Assuming it is now considered legacy.

Can ROOL not purchase the suite for a nominal fee? Assuming it is now considered legacy.

I think the answer is what Sprow said:

The blockage with all things ANT was most recently covered by Steve Revill in his London show 2018 talk. The path has become more twisted since ANT were bought out (after ROOL was formed) so it’s a lot of effort (on ROOL’s part) to keep restarting talks with another set of contacts who have no idea what they’re talking about!

Basically, the code is antique, the current owners may not even know whether they have a copy of the source and not know how to attach a value to whatever they can find anyway.

may not even know whether they have a copy of the source

Is the version we’re running not ARMv7 safe? That implies that somebody somewhere has a copy of the source.

and not know how to attach a value to whatever they can find anyway.

It’s a working but buggy implementation of a generic bit of a network stack for an operating system some thirty years old.
To put it on context, how much would you expect to ask for a “file manager” (trying to think of something useful but not unusual) for Windows 3.1?

Seriously though, how did we get to the stage of needing flippin’ NDAs for DNS lookup? Didn’t Acorn ever write a resolver?

That implies that somebody somewhere has a copy of the source.

I would assume that Acorn had a copy of the source, which passed to Pace, Castle and then ROOL.

Seriously though, how did we get to the stage of needing flippin’ NDAs for DNS lookup? Didn’t Acorn ever write a resolver?

As I understand it, that’s the whole issue. Acorn used ANT’s resolver, under a licence which gave Acorn the source to use and update.

To release the source, however, ROOL would need whoever now owns ANT’s IP to agree to a change in licence. As Sprow wrote, that entity keeps changing, and the public-facing legal bit doesn’t actually know that it owns the rights to the Resolver. In order to agree to the new licence, the relevant people therefore need to work out what the RISC OS stuff is.

It’s probably worth noting that even the specification for MBufManager (and the rest of the DCI4 spec) is covered by an NDA, so rewriting MBufManager isn’t just a matter of writing something from scratch which meets the published spec, as you’re not allowed to read the published spec!

so rewriting MBufManager isn’t just a matter of writing something from scratch which meets the published spec

So, just out of interests sake – what crashes/fails if one unplugs MBufManager? Does the Internet module use it? Could its behaviour be determined by looking at how it is being used? Or are we up you-know-where without a paddle?

So, just out of interests sake – what crashes/fails if one unplugs MBufManager?

Nothing apparent.

Does the Internet module use it?

Yes. Unplug MBufManager and type ping 8.8.8.8 see where things go.

Another experiment, leave MBuf alone and unplug resolver and ping 8.8.8.8
From that you see that if you can manage with just IP then you can work without Resolver.
In fact if you put the appropriate name —> IP items in hosts you should be able to use things like https://www.riscosopen.org/hub/

Could its behaviour be determined by looking at how it is being used?

If you’re minded to do so, reverse engineer Internet —> MBuf to see what use it makes of the MBuf facilities.

Or are we up you-know-where without a paddle?

Without coders is mostly the problem, I keep seeing these requests for people with a clue.

As to Resolver well, _GetHostByName, GetHost, _GetCache, _CacheControl all seem to be giving big clues about what it does without deep description of the method.

Scrat around on the net for a stand-alone DNS client source package and hack it to match the existing Resolver.

As to Resolver well,

Resolver isn’t that big a problem. DNS is a known entity and it replies in a standardised block.
It isn’t a black box like MBufManager.

Scrat around on the net for a stand-alone DNS client source package and hack it to match the existing Resolver.

My understanding of DNS is basically “name in, IP out”, however thirty seconds of Googling came up with something that might be useful and it has a friendly looking licence. ;-)

https://github.com/wahern/dns
https://github.com/wahern/dns/archive/master.zip

Unfortunately I didn’t understand most of the code (tea? WTF is tea in a resolver?), so this is probably best left in the hands of somebody who knows what they’re talking about.

My understanding of DNS is basically “name in, IP out”, however thirty seconds of Googling came up with something that might be useful and it has a friendly looking licence. ;-)

I heard good things about c-ares by Daniel Stenberg and friends, which is used by (lib)curl: https://c-ares.haxx.se/

Also MIT-licensed which is always a good thing.

Unfortunately I didn’t understand most of the code (tea? WTF is tea in a resolver?),

There are currently two DNS methods (old bog basic non-encrypted and newer encrypted) but you know that.
Modern name resolvers include facilities for encrypt/decrypt and in this case the author used Tiny Encrypt Algorithm TEA

It’s the way things are going if you check

In this case – black box, magic happens inside – unless there’s a problem when it’s in use just nod and look at the next bit.
Uppercase to note the Acronym status of TEA. (Like LASER, that lots of people¹ spell wrong)

¹ Particularly Americans, but then they didn’t invent it so why would they bother?)

There are currently two DNS methods […] but you know that.

I do now.

It’s the way things are going if you check

I presume that’s to stop [Russians|Chinese|whoever we hate today] from messing with DNS and bringing everything to a screeching halt?
I mean, think of the trauma if [Facebook|Instagram|Snapshat|Skype] aren’t available! People would have to… interact… with each other… it could get nasty…

In another “way things are going”, there’s IPv6 on the horizon. I don’t think it will make much impact to RISC OS for a good while yet (there’s a ridiculous amount of IPv4 only kit around that I’d imagine future routers on an IPv6 Internet would fake up some sort of NAT style arrangement for the old stuff).

Uppercase to note the Acronym status of TEA. (Like LASER, that lots of people1 spell wrong)

Shouldn’t we be pedantic and write T.E.A. and L.A.S.E.R. ? ;-)

BTW, the use of “Laser” (written like that) is a peculiarity of acronyms that are pronounced as words. You probably say something like “lay-zer” and not “ell-ay-ess-ee-are”, and this is reflected in the acronym being written in lower (or initial cap) case. Same with scuba, radar, and care package. People would think it odd if you said you went S.C.U.B.A. diving and found you received a C.A.R.E. package in your absence…

In another “way things are going”, there’s IPv6 on the horizon. I don’t think it will make much impact to RISC OS for a good while yet (there’s a ridiculous amount of IPv4 only kit around that I’d imagine future routers on an IPv6 Internet would fake up some sort of NAT style arrangement for the old stuff).

Yep, there’s definitely a lot of v4 stuff out there. My ISP supports v6 but my router doesn’t. If I plug my PC in directly, bypassing the router, then I can make a v6 connection, not that there’s much benefit in doing so.

People would think it odd if you said you went S.C.U.B.A. diving and found you received a C.A.R.E. package in your absence…

Well, I just learned that CARE is an acronym. I always took “care package” to mean something like “a package to help take care of you”. With that said, I don’t think I’ve ever heard the phrase in real life (just occasionally on the Internet).

Well, I just learned that CARE is an acronym. I always took “care package” to mean something like “a package to help take care of you”. With that said, I don’t think I’ve ever heard the phrase in real life (just occasionally on the Internet).

Likewise.

And while I knew that SCUBA is an acronym, I still don’t know what it stands for. “Underwater Breathing Apparatue” I guess, but SC? Dunno.

Self-contained. I had to look it up.