BBC News - Alpha Preview

The latest version of NewsUK v1.15 is avaliable. It’s just been released so it’s not in !Store yet.(awaiting a bit more testing)
But I’ve added a “Feeds” menu to allow other BBC Feeds to be used (Including the sports feed, which isn’t on the news.bbc.co.uk website)
You have to wait until the feed has refreshed to change the feed, or change it and refresh manually after the feed finishes loading.
Hopefully it will also fix the rare corrupted file issue, although it happens rarely it’s difficult to even reproduce.
I got bored after refreshing 60 times, and it didn’t happen during that – so maybe it is fixed…
The refresh is faster than previous versions as this has some socket and DNS tweaks in there too, which I hadn’t bothered to put in before.
It’s still quite kludgy as far as it goes, but it works and this was a nice improvement that didn’t take too long to add as I essentially merged some code from WeatherUK.

Avaliable hopefully from !Store soon, and from here

Ready for the Wakefield show Tommorow is a new version (v1.18) of NewsUK avaliable on the !Store
There are loads more feeds, and the long standing issue of selecting a new feed and not forcing a refresh is fixed.
Enjoy! Hope to see you at the show.

The last 1.xx version of NewsUK has been released, version 1.20 as always it’s avaliable via the !Store.
This version brings an option to use RamDisc, Refresh timing options and rendering options.
This was designed to work better on my RISC PC, it’s still pretty nasty on a RISC PC as for me it takes around 1 min 30 secs to load the main feed, but at least using the RAM disc and rendering in 256 colours (if your in a 256 colour mode, obviously) and refreshing every hour makes it considerably less painful.
For comparison it takes around 7 secs on my ARMiniX.
To be honest the best bet is to use Memphis3 and Set Memphis as Scrap directory and not to use the RAM Disc option in NewsUK – as NewsUK uses the Scrap directory to store image data etc. and using memphis like this effectively does the same and makes other applications (like netsurf) really fly and reduces wear on the SD cards, also it looks like Memphis is quicker than RAMDisc on a RISC PC at least.
Also copying ChangeFSI into RAM and running it’s !Boot also helps .

UPDATE: I’ve just started to have a serious issue with memphisv3 on boot where it causes filecore in use, so for the moment I’ve stopped using it – I just added it to the !Boot tasks, maybe that’s not the way to do it? :-(

Anyway hopefully you like the latest additions.
v2.x series is going to take a little while as I need to develop and test my downloading module before beginning a total rewrite of !NewsUK and then !WeatherUK – then the real fun begins ;-)

Contrary to my previous statement, v1.21 is the last v1 release ;-)
This release is just a sprite update, anyway big thanks to Richard Ashbury for the nice new shiny graphics!

As usual avaliable exclusively on the !Store

Found some seriously nasty interactions with ChangeFSI which are hopefully fixed now, so there’s a v1.22 release just been pushed to plingstore…

Version 1.23 is out, the 32bpp option causes complete destruction. Whatever is going on it’s badly broken.
It’s kind of pointless anyway as 16bpp is more than adequate, so 32bpp has been disabled.

There is another update for NewsUK on it’s way. This update has a minor (and possibly pointless) update for closing sockets.
More importantly I’ve bundled the latest version of ChangeFSI (1.52) which fixes a long standing issue on the ArminiX (Pandaboard).
Fantastic work in tracking this little pest down by ROOL

This update has a minor (and possibly pointless) update for closing sockets.

Ever had a machine on for a long while and ended up with no available socket for you news or mail client because things failed a few times when your ISP dropped the line for a while?

Cleaning up open sockets when you don’t need them anymore is always a good idea.

Cleaning up open sockets when you don’t need them anymore is always a good idea.

Having the OS clean up automatically when things haven’t been used for a while would be nice.

Having the OS clean up automatically when things haven’t been used for a while would be nice.

I hope you don’t mean ‘like Windows does’.

Having the OS clean up automatically when things haven’t been used for a while would be nice.

Having the opinion of heiki-heiki, the OS will deal with it just encourages crap coding. Why bother freeing sockets? Why bother releasing memory claims? Why bother?

Grrr…

It isn’t hard. You are handed a handle. Check it periodically if nothing interesting is happening.

The update adds shutdown before the socket gets closed, it was getting closed before – I’m just adding a shutdown call to the socket before I close it – it seems to make good sense, although I’m not sure it’s strictly needed.
Many, many years ago I had to spend a few weeks debugging a lovely memory corruption caused by a socket not being closed (and re-opened) as the person who wrote it couldn’t figure it out. It crashed every 2 days or so and in random places. Sometimes in a printf, sometimes elsewhere – it drove me nuts! As using gdb will!
Ever since those days I’ve learned the lesson, close all I/O handles as soon as is possible (within reason) and check your code at least five times that all sockets are definately closed. This doesn’t mean one may not slip through, but it reduces the chances of having the pain of debugging this kind of nightmare.

How would the OS know whether the socket is being used or not? e.g. if a network service (http etc) is running the socket needs to be open permanently and may have no traffic or usage for possibly days.

Just a quick note to say that this update is now live on !Store, and has been for the last few days.
> [intercept.transmission]
> i.am.coming
> [.]
We recommend upgrading to the latest build for the best experience, esp. if running on ARMiniX or Pandaboard.

Andrew what is happening with your posts? I noticed that intercept message appeared on your recent CSA Delegate announcement too.

Andrew what is happening with your posts? I noticed that intercept message appeared on your recent CSA Delegate announcement too.

Teaser?

It’s hard to say – my news postings and forum posts seem to be being intercepted by something at the moment. Fred might be right, but there’s still some investigating to be done.
[intercept.transmission]
i.aM.near
[.]
Anyway, enough posting – things are quite busy here preparing for the show, as you can probably imagine!

Another small update, this one prevents those dodgy UTF-8 characters from showing up (well replaces them with spaces at the moment)

Great news, thanks Malcolm! Any updates with regards to !Weather?

Thanks, just to give you an update as regards my projects.
I am making (slow) progress with my accelerator module, (it’s designed to offload the downloads into a single program, which in theory should improve performance and more importantly remove the complexity of downloading from the app).
My Jabber program is on hold until I get the module completed.
My weather application will be the first app to use the new module, as it’s the simplest of my applications.
My intention is to support:
BBC weather feed (xml)
The Met Office (xml or json)
Yahoo (xml or json)
wunderground (xml or json)
openweathermap (looks like they’ve run out of disc space! So not sure about this yet)

The UTF-8 work is being done to support xml better, hence the update to NewsUK as it already had a filter in place to deal with non-ASCII characters – it was just pretty basic, so exending it is fairly easy and allows me to test it and be useful at the same time :-D

I’ll be adding xml as well as json for feeds that support it, as I’ve had reports of xml being blocked by some filters and json is actually nicer and easier to deal with anyway!
So options will be something like “Met Office (xml)” and “Met Office (json)” etc.

The important question is timescales, one I can’t give an answer to. I would like to say I will get the first Weather update out before the end of this year, but it’s proved difficult to get much done so far this year. My target was to have my accelerator done by March, RSS done by June and Rabber done by August and then new versions of Weather and News for Sept. But I’ve not even mananged to get the accelerator done yet, quite depressing to be honest!

For me progamming is 10% skill or less and 90%+ effort on RISC OS, I assume the same is true for ROOL/ROOL contributors as well.
They are all immensely skilled IMHO, so that should give you an idea of the effort I think they put in (given it’s not their full time jobs, it should be – if only it could!)
Please note there is a nice page here
If every user gave the minimum a month, it would still be a reasonable amount.
To quote the page: “If you can’t think of a specific bounty that you’d like to put your money towards, then please consider this one. If we ever find we have a surplus here, we will distribute it evenly across all of the other bounties.”

I hope that your module will be finish soon. To be able to fetch http pages, xml and json feeds from Basic, could open the door to new applications. If BBC News becomes more generic, it could be fun also to adapt it to other feeds. An RSS kit?

Carrying on with the UTF-8 work, version 1.26 is avaliable. This adds conversion for the GBP symbol and the Euro symbol.
They seem to be the main ones that show up
Avaliable as always off the !Store

Carrying on with the UTF-8 work, version 1.26 is avaliable. This adds conversion for the GBP symbol and the Euro symbol.

Um, you’re not implementing UTF-8 to Latin 1 conversion from scratch, are you…? That’s already a solved problem for RISC OS: have a look at Iconv as used by NetSurf.

Kind of yes, but it was an exercise in understanding UTF-8 more than anything – fairly easy.
My aim (long term) is to do proper UTF-8 conversion and display.
First one will be Arabic (which will be nice as it’s right to left).
Then I fancy doing some simple character sets such as Kanji and Chinese (go on Rick, bite)
The only issue I see is that I need to do font selection based on the character set, and move between fonts depending on the character.
But that’s fun for much later when I get !RSS off the ground
I’ll have to take a look at iconv before I do any work on this

That’s already a solved problem for RISC OS: have a look at Iconv as used by NetSurf.

Iconv works fine (obviously) – I use it for decoding UTF-8 and UTF-16 encoded ID3 tags in my ID3TagEd application.

(go on Rick,bite)

Hiya!

I’d suggest work on kana or Hangul first, something that can be tested without the additional complications of right to left (not to mention the complexities of Arabic script!).

and move between fonts depending on the character.

Really, this is a job for FontManager, instead of everybody inventing the same wheel over and over.