Wiresalmon
Simon Willcocks (1499) 509 posts |
Has it stopped working? Presumably it was 32-bit. It shouldn’t corrode over time. |
Steve Pampling (1551) 8154 posts |
Welcome to the edge of the reality that all people dealing with medical equipment suppliers1 have to delve into (ugh, yuk) I found that L2 / L3 differentiation is, apparently, a difficult concept to grasp. Whatever you do, don’t start on an explanation of how application layer on device 1 communicates to application layer on device 2… :) 1 Who still supplies equipment with dependencies on broadcast for critical data? |
Rick Murray (539) 13806 posts |
Seventeen years. So that’s ARMv6 (supported rotated loads), early (permissive) ARMv7, later (strict) ARMv7, the no-SWP ARMv8… Standing still is a form of corrosion. Just ask any statue. |
Steve Fryatt (216) 2103 posts |
Did you actually read my post? There’s a vast difference between running the existing Sunfish code through a modern compiler that knows of ARMv7 and ARMv8, and possibly fixing a few null pointer dereferences along the way, compared to writing a new front-end from scratch. As it stands, Sunfish is functional for my needs: the filing system works, and can be configured by editing text-based files. The filing system can be re-built already, as Chris Gransden has shown. The filer is crashy if one goes too far beyond mounting shares and opening directory windows, however, and has been post-Iyonix. If that can be fixed with a re-compile, that’s good and I might be prepared to invest that time in it. If we’re talking of writing a new filer, however, then I’d need to think a lot harder about the cost/benefit analysis. IIRC, I wasn’t able to pursue this when I last looked because of the missing library and some funky compiler setup. The second one may be solvable (if I can even remember what the details were), if all of the code is available.
All of which can be fixed by a quick re-compile, unless there’s hand-written ARM assembler in there. If I’m correct that it’s a C++ library, then I’ll pin my hopes on the idea that the type of person writing one of those probably wasn’t that much in to hand-written ARM assembler — at least until I get time to look at the code (which won’t be soon). |
Dave Higton (1515) 3497 posts |
Yes. I’m astonished that you could think otherwise. I’m not a fan of user-generated libraries that have been abandoned many years ago. Yes, the source is available; yes, I’m (almost) sure that I could recompile it and have a working library. The small reservation is that it used code features that weren’t available in the DDE then – of course that may have been overtaken by time. I’m well advanced with a new front end app for Wiresalmon, written in C using the DDE and Toolbox. The Wiresalmon front end is functionally quite simple. I haven’t looked at the Sunfish and Moonfish apps in some years, and I didn’t go into them in much depth when I did, so they might be considerably more complex – I don’t know. |
Rick Murray (539) 13806 posts |
Given the woeful state of “official” libraries (RISC_OSLib is horrid, OSLib is a wrapper around the API, and they tried to do better with the Toolbox but I think I’d was too little too late) it’s not a surprise that there are numerous bits of third party library code floating around. And, yes, in varying states of disrepair.
That’s probably the best way. Unpicking old third party code with “dependencies” can sometimes be more trouble than “sod it, I’ll just roll my own”. |
Steve Fryatt (216) 2103 posts |
It was probably something to do with the fact that you responded to a simple question about the whereabouts of a library needed to rebuild an existing, complete and largely working application with a lecture on how to select a library when starting a new project from scratch. |
Dave Higton (1515) 3497 posts |
I told you where the source is. I expressed a couple of opinions, making clear that your choice was yours. No lecture. |
Dave Higton (1515) 3497 posts |
I’ve just got my new Wiresalmon front end, and my new build of the Wiresalmon module, capturing, apparently successfully. Alex’s original was released under the GPL, and my version will be too. I need to do a bit more testing first though. |
Dave Higton (1515) 3497 posts |
Does anyone have any feeling of whether the pcapng format is any more popular than original pcap? If it is, that would be a logical upgrade to the Wiresalmon module. |
Rick Murray (539) 13806 posts |
If it isn’t, it may well be… |
Colin Ferris (399) 1809 posts |
Would be handy for there to be a book or PDF on how the Internet capture works. Err in simple terms :-) |
Rick Murray (539) 13806 posts |
Maybe there is? Sounds like the sort of thing O’Reilly might do, complete with a drawing of a random animal on the front. |
Theo Markettos (89) 919 posts |
Oh, that’s embarrassing. That was really just a dump of the SVN from Graham Shaw’s server in order to build something, but it seems that server has gone now. However the wayback machine does confirm the github above is the most recent commit (from 2012) up to 2020, and I doubt anything happened since then. This is indeed a problem with one-person libraries, although Graham’s stuff is fairly well engineered (better than the RISC OS average, TBH) so I wouldn’t have too many worries about reviving it. Whether it’s worth doing so depends on whoever developing wants to do. |
Steffen Huber (91) 1949 posts |
Graham also wrote a proper book about how to use RTK, so compared to other offerings, it was also rather well-documented – but I fear that the book might now be lost forever, since the only way to get it was to order via his website. |
Theo Markettos (89) 919 posts |
It seems like Graham is still about and has a contact email if anyone wants to get in touch and ask him. That’s assuming the @riscpkg.org email still works, which appears to be in the DNS and is listening to SMTP, announcing itself from one of his other domains. |
David Pitt (9872) 362 posts |
|
Theo Markettos (89) 919 posts |
That’s the one. I didn’t know it had an ISBN (9780954796204). It looks like Google let you read the whole thing. |
Dave Higton (1515) 3497 posts |
I discovered that the module has to have alignment exceptions disabled, otherwise the machine stiffs completely (no cursor movement, no response to Num Lock). I wouldn’t have known before because the app crashed immediately if they were on, so the module always ended up running with them off. |
Colin Ferris (399) 1809 posts |
You seem to have rebuilt the C side of the module – did you check the machine code file that I presume is part of the module? |
Dave Higton (1515) 3497 posts |
I just assembled the assembly language, compiled the C, and linked the two. I did change one ADR to a MOV (I think that’s further up this thread) but that’s all. I have hardly used assembly language in many years now, so checking it is going to be difficult. I’m going to have to go through all the code to try to track the problem down. When I wrote the two printer dumper modules, I made extensive use of Reporter, and I expect to do the same here. I would be grateful to anyone else for checking the code. |
Colin Ferris (399) 1809 posts |
I would use Druck prog to disassemble the mod and see if there was anything that was obvious wrong. Zap is also handy for looking at modules. Locking up – interesting:-) Is there a download – only a Vrpc emulator here at the mo. |
Dave Higton (1515) 3497 posts |
There is a download of the module in a TEMPORARY LOCATION – not any more. This will disappear at some point in the not too distant future – it has, see further down the thread. But I’ve replaced it with a full download of app, module and sources. |
Rick Murray (539) 13806 posts |
It’s a module. So if it barfs in a privileged mode, oops. |
Dave Higton (1515) 3497 posts |
Hmm, around line 403:
Those two LDR operations can’t both be word aligned, can they? Not sure either why it’s loading a word and storing a byte in each case. |