fylfot
29/4/04 9:31PM |
Let's hope for a roadmap for RISC OS 5 development very soon! |
tribbles2
29/4/04 9:46PM |
The thing I'm not entirely happy about is that you can only subscribe if you bought it at least 3 years ago, and you have to prove when you bought it - I can't remember when I bought my copy: don't they have records?
If you can't prove it, you've got to buy a new copy... |
philipnet
29/4/04 10:13PM |
In reply to tribbles2:
The wording on Castle's site clearly states (well implies  ) that the requirement of prove is only applicable if order on-line.
There's no such working on the PDF order form.
--
I am trapped on a roof ... with an unstable human who drinks too much whiskey and who called me a smurf.
Angel - Underneath (Season 5, Episode 17) |
dgs
29/4/04 11:05PM |
It's good to see Castle actually plan out future compiler developments for RISC OS 5 and beyond, and how people will pay for them. RISC OS users increasingly have to realise that future development depends on investment by the user base. Initiatives like this, just like Stuart Tyrrell's scheme for paid support for USB, provide the funding to ensure that development continues.
Developers may think that they are entitled to the whole thing for free....
"I want it for half the price", of course.
dgs |
md0u80c9
30/4/04 12:36AM |
Sounds promising - nice to see the compiler in active development |
tribbles2
30/4/04 8:44AM |
In reply to philipnet:
"Subscription only
To qualify for the subscription prices, purchasers will have to provide evidence that they have purchased a full copy of Castle's C/C++ development tools when registering for upgrades."
About half way down on the first page...
Thing is, I think it was just over 3 years ago that I got it  |
philipnet
30/4/04 9:16AM |
In reply to tribbles2:
The full text for point no. 1 is:
"To qualify for the subscription prices, purchasers will have to provide evidence that they have purchased a full copy of Castle's C/C++ development tools, within the last three years, when registering online for upgrades."
So I want to know what evidence will be suitable?
I doubt that I can pull out my receipt for the tools from the 2002 Guildford show. Will my serial number do?
--
I am trapped on a roof ... with an unstable human who drinks too much whiskey and who called me a smurf.
Angel - Underneath (Season 5, Episode 17) |
tribbles2
30/4/04 9:20AM |
In reply to philipnet:
Damn Adobe PDF Reader not selecting the right text!
I ordered mine over the 'phone from them, so I suppose mine should be easier if they had to prove it.
People who bought at a show (like yourself) will be harder, as they (probably) won't have customer records including addresses and so forth. |
bernie
30/4/04 10:07AM |
C Compiler Select?! A sort of dejá-vu.... |
jbyrne
30/4/04 10:34AM |
Given that Castle only released the 32 bit tools 18 months ago, the three year restriction isn't going to matter for another year and a half though. As long as you can prove that you have a legitimate copy (i.e. you have the original CD and serial number) surely that will do. |
mrchocky
30/4/04 10:49AM |
Well, it's a nice idea, and I understand that Castle have to fund continued development of the tools.
I purchased the 32-bit tools, but I won't however bee subscribing to this. As Chris has mentioned, GCC already does pretty much all the improvements mentioned (some such as inline assembler for many years), and GCC 3.3 knows already quite a bit about XScale optimisations (and I'm now experimenting with having XScale tuning the default, which will increase performance on StrongARM too), not to mention the ongoing efforts to improve overall ARM performance on other ARM platforms, which are relevant.
The only notable exception remains modules[1], and the debugging features supported by DDT, and of course GCC not being the fastest of compilers.
[1]. Jeffrey Lee has an experimental version of LCC that claims to do modules. |
Phlamethrower
30/4/04 2:33PM |
"[1]. Jeffrey Lee has an experimental version of LCC that claims to do modules."
Indeed I do. And if anyone has some modules they've written for Acorn/Castle C/C++, then I'd be very grateful if they could get in contact so they can help me with testing out LCC |
Spick (-1.0)
1/5/04 1:38PM |
C compiler it is, C++ compiler it is not. |
philipnet
1/5/04 7:47PM |
With all this talk of C, I hope they've also fixed the niggle(s) I've had with ObjAsm.
--
I am trapped on a roof ... with an unstable human who drinks too much whiskey and who called me a smurf.
Angel - Underneath (Season 5, Episode 17) |
adrianl 
2/5/04 3:24AM |
"niggle(s) I've had with ObjAsm"
Bugs or just things you dislike? I make heavy use of ObjAsm and I'm not aware of any bugs in Castle's latest version. |
philipnet
4/5/04 11:29AM |
I think it's just one niggle I've had (which affects OSLib 6.60 and later users), and that's the -I argument. It doesn't work like -I does for the C compiler.
I.e. Specifying -IOSlib: won't expand OSLib: the way that the C compiler does and ObjAsm won't then search OSLib: for the Hdr files I GET. To work around that you have to issue something like:
do objasm -I OSLib:
Note the 'do' and the space after -I. I don't like this inelegant solution and so I've stuck with OSLib 6.50 .
Note: OSLib 6.60+ has switched from absolute filenames to 'relative' ones for it's GETs.
--
/Some day I’ll go where/There ain’t no rain or snow/Till then I’ll travel alone/
/And I’ll make my bed/With the stars above my head/And dream of a place called home/
Angel - Shells (Season 5, Episode 16) |
| |
