When will the Iris web browser be released to mere mortals

Does anyone know what the schedule is?

There have been hints it will be available to those who purchased !OBrowser to support the development soon. Then more generally thereafter.

But it might be worth watching the Virtual Wakefield Show next Saturady 24th. Apparently I understand RISCOS Developments have lots to announce/discuss etc. The show will be on ZOOM, streamed live on the Club’s YouTube Channel and hopefully live on Facebook as well. Full details on the show website.

https://www.wakefieldshow.org.uk

I bit the bullet, and bought !OBrowser on !Store. This indeed gives access to the !Iris beta.

Thanks for your support, Charlotte – enjoy the software. I know it is a weird way to make it available, but we wanted to stress that Iris is not a complete/polished/ finished project, and is very much alpha/beta still. It just happens that it can already play a useful role in the RISC OS armoury, if with a lot of rough edges.

I bit the bullet

I bit the bullet too. And WOW!!! So it’s a “Hello from Iris”.

It’s impressive. (But it is beta still.)

I’m all in favour of “make it work first, then make it work fast”, but trying to log on to my bank with Iris was painful. Typing into text fields seems to only accept one character per second. Even I type faster than that.

I’d suggest that is an area in serious need of work please.

I’d suggest that is an area in serious need of work please.

Given that my lowly “freebie when you subscribe to a magazine” tablet can run Firefox at a reasonable rate (given all my blocking add-ons) on hardware probably akin to a Pi3, and I don’t need to own Iris to know it’d have it’s ass handed to it by Firefox…

…I can’t help but think the operating system itself might be part of the issue. Let’s see – it only uses one core, it doesn’t understand threads, it’s a cooperative multitasker, there’s no GPU assistance with rendering…shall I go on?

I’d like to give Iris a whirl myself to see how it fares (especially on my site), though I don’t use RISC OS for much network stuff because.

I’m surprised and impressed if it can cope with Google Docs, but I’m even more surprised and impressed that it works at all given how alien RISC OS is to everything else out there.

Typing into text fields seems to only accept one character per second.

Bank URL? I’m guessing there’s a pile of JavaScript to handle input. But the question is, are they using custom routines for what they need, or are they pulling in the entirely of big libraries like JQuery and that other one in order to handle something in three hundred lines of code when twenty would do?

I’m all in favour of “make it work first, then make it work fast”, but trying to log on to my bank with Iris was painful. Typing into text fields seems to only accept one character per second. Even I type faster than that.

If take a look at Tasks Window (application Tasks and Dynamic areas) be ready for a shock as to how much memory is be used, including Shared Library, I think on my computer it came to near a Gb. So somethings like typing text will likely be slow.

Guys, please lets not complain about Iris at this stage.

I did my support (even if I had previous access to it because of the i.mx6), because RISC OS really need a modern browser and because such type of products will take an impressive amount of work to get done and work decently on an OS that is not designed for modern workloads and programming paradigms.

As Rick mentioned well, it’s mostly the OS (An Acorn MOS pumped with some steroids without a modern architecture and with lack of hardware acceleration support, a multi-tasking that require a lot of work to make an app responsive and everything executed on a single core) that is causing most os the slowness on Iris. WebKit is designed for modern, preemptive, multi-core OS with good hardware acceleration support.

Also let’s not forget that even on more modern OSes, browsers struggled a lot on a Raspberry Pi to become acceptable and there are still a lot of issues (on linux and other OS), so please let’s not develop perceptions that might be unrealistic…

Note to Alan about speed – I’m not sure what machine you’re running on, but I have noticed that text input seems to benefit noticeably from faster hardware. On my my main ARMX6 RISC OS 5 work machine, it isn’t great. On my 4te, it is considerably better. Considerably. (apologies for the Bond mis-quote).

It is certainly an area of performance that we’d like to work on. It will take some digging to find just where the bottlenecks are, as many text boxes also trigger JavaScript for dynamic processing of what you’re typing. So, you can have text input, JS, plotting, GTK layer, text shaping manager (harfbuzz is HUGE – 10-20MB library) and so on all layered on top of each other. Finding the happy medium and working out priorties will, I think, be key, but not easy work.

I know that Ben Avison of ROOL did some significant work in this area for the Pi Foundation when they improved their browser a few years back (I remember seeing Ben’s name in the credits), so as and when he’s available for contract RISC OS work, perhaps we can hire him to do a similar pass over Iris. We’ll see. In the mean time, Lee’s doing a pretty amazing job, and I don’t want to take any credit away from him. He’s worked miracles, he really has.

I think it’s worth reiterating what an incredible feat it is to have a genuinely modern WebKit browser on RISC OS. It’s a long way from perfect but it’s light years ahead of anything else on the platform and given a sufficiently fast computer (such as the Pi 4), it is actually a reasonably good experience. That might sound like damning with faint praise, but it is remarkable that a “reasonably good web experience”, relative to other OSs, can be had on RISC OS. And we’re not even at the final release yet. Thanks to Lee, Andrew and all who contributed to make it happen.

Indeed! It’s ‘usable’ on my Pi 3, and I imagine that it must be better on a Pi 4. I need to find a case for my Pi 4 as it’s just sitting in storage at the moment due to not having a case!

I’m not sure whether to report bugs at this stage as I’m sure there’s a huge list already.

Thanks to Lee, Andrew and all who contributed to make it happen.

I would like to add my thanks as well. Iris works very well on my Pi4 clocked at 2GHz. It may be a work in progress but what has been achieved so far is amazing. Andrew keep up the good work and let us all know how you manage to squeeze so much into your day.

Thanks to Lee, Andrew and all who contributed to make it happen.

Absolutely! and yes on more powerful boards it seems to be working smooth (I tried it on my overclocked RPi400 and it works really smooth)… Writing an article for my blog now with some screenshots and a picture that shows first ever compiled C to AArch64 on RISC OS using Iris Browser :D (just using Compiler Explorer but it works!)

Also managed to use Codeanywhere IDE on RISC OS using Iris, it works ok, the editing is smooth and most of the features seems to be working (it needed some tweak however, so not production ready).

Obviously many modern websites like Amazon, Twitter etc… seems to be working… I wonder if I can run Koding on Iris and use it to control GCCSDK… another project for the weekends I guess :)

I hope that the offer in !Store of OBrowser/Iris stays until i got my next salary (24th of may).
At the moment sadly i just can not afford anything.

It works very nicely, except on sites that are absolutely choked up with Javascript, where it’s substantially slower than Chromium on the same hardware. (This is presumably because Chromium can make use of multiple cores and other advantages that come with Linux.)

The only substantial bug I’ve noticed is in image upload. I tried to Tweet a PNG, but Twitter didn’t recognize the image.

It doesn’t work very well for a Teams meeting.

I think you got those words in slightly the wrong order:

Teams, it doesn’t work very well for a meeting

:-)

Teams, it doesn’t work very well for a meeting

Say hello to Teams, say goodbye to most of your spare RAM.
Like most MS stuff it will probably be quite usable when it comes out of beta.
Somewhere round 2032 should be decent, of course the competition will have far surpassed that level by that date.

Little Iris tip – if you have the spare RAM (suggest about 300MB), create/configure a RAM disc and copy Iris into it. Load times are noticeably quicker, as are some aspects of page rendering (due to font system, I think).

I was just trying out more comprehensive unicode support by adding UniFont ttf file to Iris (about 30MB of fonts), and there was a noticeable negative hit from usual storage media (Wikipedia in chinese took about 30 seconds to appear), whereas using RAM disc there was no real hit at all (3-4 seconds). It is much less noticeable with normal typefaces, and probably an area we can optimise, but if you’re making significant use of the browser, it is worth noting this benefit.

Is anyone trying the latest versions (JIT enabled) of Iris with Alignment Exceptions on?

I made some tests: JavaScript is on par with the Raspberry Pi OS browsers (1 core of course) and Chrome on Core i3.

Problems:

• Disk I/O (soon to be solved with Sata disc support under the Compute module 4).
• Graphic stack: would really need some acceleration.
• PMT: a full screen mode with more aggressive “harvesting” of resources would help.

Is anyone trying the latest versions (JIT enabled) of Iris with Alignment Exceptions on?

I normally always have AEs on. However, Iris is unstable when run with AEs on. On the other hand, with AEs off, I have only had one crash in three days use.

Re Alignment Exceptions

This is explained in the ReadMe file in the ZIP archive.
Quote:
In order to be able to track real-world developments, Iris is (as
you have read above) based on the cross-platform WebKit project.
Some elements of Webkit, most notably the half-word Thumb2
JavaScript JIT (Just In Time) compiler, require alignment
exceptions to be disabled to function correctly.
End Quote:

Its to do with needing to use Thump2 (16bit) to give us some of useful features.

On my ARMX6 with a ‘High’ ROM V5.29, and Alignment Exceptions on this latest version of Iris stopped working after trying to get a link on the current web page to open in a new tab.

Since Iris is practical only on computers with recent ARM v7 and later chip designs, running Iris with Alignment Exceptions off could corrupt Iris data or other data. Yet we are advised to run Iris with Alignment Exceptions off.

It would have made more sense for the Iris code to have been created so as to work with Alignment Exceptions off just as most other applications have been made to work on modern processors; but, maybe, there are technical issues why this is impossible?