Re: [Rpcemu] Future potential features

Some feedback from me.

> Peter Howkins <rpcemu.howkins@marutan.net> wrote
[snip]
> 1) A dynamic recompiler targeting the ARM64 instruction set. Whilst at an
> early stage this already shows an impressive speed gain over the
> interpreter version on the same hardware (at the show we used a
> Raspberry Pi 4 with 64bit Linux on).

This is certainly interesting, and would be very helpful thinking
about a possible future with a working Linux port on Mac hardware.

On the other hand, for the use case "provide compatibility for old
software on RISC OS", an ARM32 Dynarec would be more important for
me. Because ARM64 is so different to ARM32, I guess the "commom
ground" is near nil.

> 2) A multi-machine configuration and launcher program. This is ideal for
> people who have multiple copies of RISC OS installed, e.g. for testing
> software under different versions.

I like the Arculator v2 implementation of such a feature, at least
wrt the amount of details you can configure. UI-wise, I would like
something different, but have not thought about it too deeply :-)

VirtualBox has also a powerful machine configuration UI, maybe some
inspiration could be taken from there. Your "New" icon in the screenshot
reminds me of that.

My typical use case is indeed "testing with various OS versions",
but because of dependencies between !Boot and OS and CMOS, I keep
the installs completely separated. Which would not really be a
problem if there were "multiple mounts" in HostFS which would make
it easy to keep "System" stuff away from the "Programs and Data"
stuff.

Before providing a full-fledged machine configuration UI, maybe
a minimal approach would be a nice and easy first step. I always
thought that RedSquirrel's "Models" approach worked fine for
98% of the use cases, bundling machine/OS-specific configs together
with CMOS config, referring to "RomSets" for the OS itself. If there
are multiple Dirs found in "Models", a launcher frontend provides
you with a possibility to select things, and if only one Dir is there,
the emulator starts directly.

Have fun
Steffen aka hubersn

--
Steffen Huber LambdaComm System – Welcome to Trollinger Country
steffen@huber-net.de
Private homepage https://www.huber-net.de/ (http://www.huber-net.de/)
RISC OS Blog http://riscosblog.huber-net.de/

_______________________________________________
RPCEmu mailing list
RPCEmu@riscos.info
http://www.riscos.info/cgi-bin/mailman/listinfo/rpcemu

Request for clearing up what a cURL handle going bad means

From 957b8087e2a6c7dcb2183d60068913a7f89b1e4f Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=C3=89rico=20Nogueira?= <erico.erc@gmail.com>
Date: Sun, 31 Oct 2021 01:39:25 -0300
Subject: [PATCH] fetches/curl: add warning when setting curl option fails

Helps in debugging the causes for such failures.

Re-organizing brackets was necessary here so the NSLOG call could fit
cleanly, and none of the code relied on SETOPT being its own block.
---
content/fetchers/curl.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/content/fetchers/curl.c b/content/fetchers/curl.c
index d36f44c09..acc5e7eef 100644
--- a/content/fetchers/curl.c
+++ b/content/fetchers/curl.c
@@ -878,9 +878,10 @@ static CURLcode fetch_curl_set_options(struct curl_fetch_info *f)
const char *auth;

#undef SETOPT
-#define SETOPT(option, value) { \
+#define SETOPT(option, value) \
code = curl_easy_setopt(f->curl_handle, option, value); \
- if (code != CURLE_OK) \
+ if (code != CURLE_OK) { \
+ NSLOG(netsurf, WARNING, "Setting cURL option " #option " failed."); \
return code; \
}

--
2.33.1

Hi!

I was recently exploring building a slightly smaller libcurl, and had
disabled cookie support in it. This made it so netsurf just looped
eternally while trying to fetch a page, because fetch_curl_set_options()
would always return an error (due to SETOPT(CURLOPT_COOKIE,...)
failing), leading to fetch_curl_initiate_fetch() always invalidating the
curl handle. In what situation can this error happen that it shouldn't
be considered a fatal one? I think netsurf requiring cookie support at
all is reasonable, but the behavior was a bit too broken, I think.

For the debugging, I have attached the change that helped me and I think
could be merged into netsurf.

Thanks,
Érico

[Rpcemu] New 0.9.4 Mac binary release

Hello all

I've created new Mac binaries following the 0.9.4 release of RPCEmu.  These can be downloaded from the usual place:

https://github.com/Septercius/rpcemu-dev/releases

The interpreter builds are universal binaries, for both Intel and Apple Silicon (ARM) Macs.

The patch itself (for those who want to compile themselves) is available here: 

https://github.com/Septercius/rpcemu-patches/tree/master/0.9.4

As ever, bugs/comments/suggestions are welcome.

Tim

Re: [Rpcemu] RPCEmu 0.9.4

> "Filing system HostFS: Must be given a name 640".

Footnote to it all...

While I still have no idea what the above is, I no longer care... :-/

Modified my 'modus renovata' and now I have all my various RISC OS version
test rigs of RPCEmu updated to 0.9.4

Dave

--

Dave Triffid

_______________________________________________
RPCEmu mailing list
RPCEmu@riscos.info
http://www.riscos.info/cgi-bin/mailman/listinfo/rpcemu

Re: [Rpcemu] RPCEmu 0.9.4

In article <5984bc8a1erinfo@avisoft.f9.co.uk>,
Martin <rinfo@avisoft.f9.co.uk> wrote:
> In article <5984b72a3fdave@triffid.co.uk>,
> Dave <dave@triffid.co.uk> wrote:
> > In article <5984b00588dave@triffid.co.uk>,
> > Dave <dave@triffid.co.uk> wrote:
> > > I guess as it has been sorted by a new install version, I should
> > > just forget it and move on. :-)

> > Created an install to 0.9.4 of my my ROD 5.28 test rig, Pah! the
> > (Builders language) thing runs with the error "Filing system
> > HostFS: Must be given a name 640".

> The *syslog boot show command does not work on RO5 - although the
> syslog module from Doggysoft does work for other uses. And is used by
> many programs for logging.

> However, If Reporter is started at boot it can show all commands and
> errors from boot.

> But you may not feel it is worth pursuing.

Thanks for your thoughts Martin, appreciated.

Dave


Apologies for the little grump yesterday, unfortunately (medical reasons)
the clock change causes me difficulties+. I'll try and behave... ;-)

FWIW. (Not much) Later before hitting the sack, I tried the process again,
this time no error presented and the RPCEmu with 5.28 worked without
problems.

Today with a fresh brain I'll run through the process again and see what
crops up. :-) (I have a vague idea to pursue...)

D.



D.

--

Dave Triffid

_______________________________________________
RPCEmu mailing list
RPCEmu@riscos.info
http://www.riscos.info/cgi-bin/mailman/listinfo/rpcemu

[Rpcemu] Future potential features

areAs mentioned in my previous Show email, we have a couple on things we wanted
to share with people. We demonstrated them at the show, but for those that
couldn't attend, here are the details. These are both work-in-progress
features.

1) A dynamic recompiler targeting the ARM64 instruction set. Whilst at an
   early stage this already shows an impressive speed gain over the
   interpreter version on the same hardware (at the show we used a
   Raspberry Pi 4 with 64bit Linux on).

2) A multi-machine configuration and launcher program. This is ideal for
   people who have multiple copies of RISC OS installed, e.g. for testing
   software under different versions.

   Here is a screenshot:
   https://www.marutan.net/rpcemu/i/romultigui.png

Both these items are unfinished but are being worked on, we don't have any
details as to when they will be completed or which release they will be in.
As things progress further it may help us to ask you to help with the
testing of these items, we will post here again at that time.

Matthew Howkins
Peter Howkins

Re: [Rpcemu] RPCEmu 0.9.4

In article <5984b72a3fdave@triffid.co.uk>,
Dave <dave@triffid.co.uk> wrote:
> In article <5984b00588dave@triffid.co.uk>,
> Dave <dave@triffid.co.uk> wrote:
> > I guess as it has been sorted by a new install version, I should
> > just forget it and move on. :-)

> Created an install to 0.9.4 of my my ROD 5.28 test rig, Pah! the
> (Builders language) thing runs with the error "Filing system
> HostFS: Must be given a name 640".

The *syslog boot show command does not work on RO5 - although the
syslog module from Doggysoft does work for other uses. And is used by
many programs for logging.

However, If Reporter is started at boot it can show all commands and
errors from boot.

But you may not feel it is worth pursuing.

--
Martin Avison using a TiMachine running RISC OS 5
and the Pluto mail and newsreader

_______________________________________________
RPCEmu mailing list
RPCEmu@riscos.info
http://www.riscos.info/cgi-bin/mailman/listinfo/rpcemu

Re: [Rpcemu] RPCEmu 0.9.4

In article <5984b00588dave@triffid.co.uk>,
Dave <dave@triffid.co.uk> wrote:
> I guess as it has been sorted by a new install version, I should just
> forget it and move on. :-)

> D.

I should be so lucky. :-(

Created an install to 0.9.4 of my my ROD 5.28 test rig, Pah! the (Builders
language) thing runs with the error "Filing system HostFS: Must be given a
name 640".

Enough... (Builders language enough) for today...

Dave

--

Dave Triffid

_______________________________________________
RPCEmu mailing list
RPCEmu@riscos.info
http://www.riscos.info/cgi-bin/mailman/listinfo/rpcemu

Re: [Rpcemu] RPCEmu 0.9.4

In article <59847ca2bdrinfo@avisoft.f9.co.uk>,
Martin <rinfo@avisoft.f9.co.uk> wrote:
> In article <59847a1611dave@triffid.co.uk>,
> Dave <dave@triffid.co.uk> wrote:
> > Even though the problem is now fixed, I really would like to know
> > what "Filing system HostFS: Must be given a name 640" means?

> Maybe it is produced by some BASIC program that is run, and 640 is the
> line number?

Indeed...
Unfortunately it happens early in the Startup Boot sequence, and as ROD
5.nn doesn't respond to CMD '*syslog boot show' I don't know how to see
what things are being done during booting.

Dave

I guess as it has been sorted by a new install version, I should just
forget it and move on. :-)

D.

--

Dave Triffid

_______________________________________________
RPCEmu mailing list
RPCEmu@riscos.info
http://www.riscos.info/cgi-bin/mailman/listinfo/rpcemu

Re: [Rpcemu] RPCEmu 0.9.4

In article <59847a1611dave@triffid.co.uk>,
Dave <dave@triffid.co.uk> wrote:
> Even though the problem is now fixed, I really would like to know
> what "Filing system HostFS: Must be given a name 640" means?

Maybe it is produced by some BASIC program that is run, and 640 is the
line number?

--
Martin Avison using a TiMachine running RISC OS 5
and the Pluto mail and newsreader

_______________________________________________
RPCEmu mailing list
RPCEmu@riscos.info
http://www.riscos.info/cgi-bin/mailman/listinfo/rpcemu

Re: [Rpcemu] RPCEmu 0.9.4

In article <5984703a32dave@triffid.co.uk>,
Dave <dave@triffid.co.uk> wrote:
[Snippy]

> While I still don't know the answer to the "Filing system HostFS: Must be
> given a name 640", I have remembered the command for the boot list.

> *syslog boot show

> Which works with RISC OS Select 6.20 versions, but does not with ROD RISC
> OS versions, reporting "File Syslog not found."

> Now I guess someone is going to say, the RISC OS 5.nn side doesn't have
> a syslog app?

> Dave

So... More work done.

As RISC OS 6.20 and 4.39 both work okay with RPCEmu 0.9.4
And...
Though I have no idea what the message "Filing system HostFS: Must be
given a name 640" means, it did set me thinking...

To that end I created another install of RPCEmu 0.9.4 using both ROD 5.28
and 5.29 ROMs (Obviously separately) :-)

This time both work correctly without the aforementioned error.

So it appears the glitch was just one of those things, or maybe something
I'd done when installing the original.

Even though the problem is now fixed, I really would like to know what
"Filing system HostFS: Must be given a name 640" means?

Dave

--

Dave Triffid

_______________________________________________
RPCEmu mailing list
RPCEmu@riscos.info
http://www.riscos.info/cgi-bin/mailman/listinfo/rpcemu

Re: [Rpcemu] RPCEmu 0.9.4

In article <59841e644ddave@triffid.co.uk>,
Dave <dave@triffid.co.uk> wrote:
> In article
> <CANc9Z=xbXjOyziiND38ooP9+Pipi+dZZJyWNaaR5q98rKXi5YA@mail.gmail.com>,
> Peter Howkins <rpcemu.howkins@marutan.net> wrote:
> > A new version of RPCEmu is available, 0.9.4
> > http://www.marutan.net/rpcemu/

> [Snip]

> Excellent news as usual. Thanks folks :-)

> Small problem and I wonder what it means...

> RPCEmu 0.9.4 ROD 5.28

> "Filing system HostFS: Must be given a name 640" presents during the
> boot sequence, and after I click it away the boot sequence continues...
> but some of the configs are missing.

> To make sure, I have done the CLI
> *configure filesystem hostfs
> *configure boot

> But that doesn't resolve it.

> Any thoughts please?

> Thanks

> Dave

> 1) This does not happen with another new RPCEmu 0.9.4 and RISC OS 6.20
> which boots and run without any problems.

> 2) I used to know a command that would show the boot list after the
> emulator was run, but damned if I can remember it, anyone help with that
> also?

> Thanks Dave

While I still don't know the answer to the "Filing system HostFS: Must be
given a name 640", I have remembered the command for the boot list.

*syslog boot show

Which works with RISC OS Select 6.20 versions, but does not with ROD RISC
OS versions, reporting "File Syslog not found."

Now I guess someone is going to say, the RISC OS 5.nn side doesn't have a
syslog app?

Dave

--

Dave Triffid

_______________________________________________
RPCEmu mailing list
RPCEmu@riscos.info
http://www.riscos.info/cgi-bin/mailman/listinfo/rpcemu