I'm not saying anything that has not been said before but, some of you may not have read it before, and I hope you might find this both interesting and useful.
Since TopModel is (still) listed as a future project, I decided to email them.
I know that most of you have more than enough to do, and so have I. But, TopModel is an application I should really like to support, and I would be interested to hear from any of you who feel the same.
Thinking of the analogy of chains, for a moment, the problem with them is that one very weak link makes all the others useless. An application may be excellent, but if just one important aspect seems very difficult - to me, at least - then that application is useless. You have got to be able to use the whole thing with acceptable ease and consistency.
With TopModle2, over the years, I have been struggling with surface rendering. I think that I have cracked it, but the results seem inconsistent and the steps complex. The apparent inconsistencies, and the complex steps, are probably interrelated. You have only to omit, or modify, one vital step and (obviously) you get a different result. However, if the differences are subtle, and not obvious and don't "shout" at you, they are not so easy to spot.
David Ruck has tried to help, and I have his article, that I have studied. But David R is a highly skilled programmer, far in advance of me. He helped me rescue my crashed hard disc. Then another highly skilled programmer friend pointed out that one of the problems with TopModel2 is that it is a "programmer's program", not all that easy.
But I want to persevere and break through.
Then, coming to the RISCOS world and returning to the "chain". It may be, that RISCOS could hold it's own, and even make progress, if there were less commercial competition within its folds and more "commercial" co-operation. It's not easy to see how this would happen but, if the many "weak links" of the RISCOS world could be strengthened in a co-operative way, we could all benefit, include those talented people who make part of their living from it.
Some of the "weak links" are where a key flagship application is either missing, or has problems of difficulty in use. I realise that there is a chicken and egg situation here. If the RISCOS world is small, developers jumps ship in order to make a living, so the RISCOS world grows smaller.
The two musical brothers, who produced Sibelius, soon jumped ship, for sound financial reasons, so did Computer concepts, and several others. Sadly, other companies, like Cerilica, have disappeared. But people like David Snell of ProCAD etc., have maintained their loyalty towards the RISCOS world, so have many others, too numerous to mention.
The problem of the idea of "commercial" co-operation is that it is almost an oxymoron (comparatively new word for me ! :-) ). How do you decide drop the effort on one of more competing applications in order to make a fewer no. excellent ? And would the originator want to share it ? Take the example of ProCAD v. RiscCAD (is it ?).
But I think that this radical, difficult, brave and selfless approach could be one of the several "shots in the arm" that we RISCOS types all need.
John N.
-- From Glorious Gloucestershire, near Lydney, using :------------ _ _________________________________________ / \._._ |_ _ _ /' Orpheus Internet Services \_/| |_)| |(/_|_|_> / 'Internet for Everyone' _______ | ___________./ http://www.orpheusinternet.co.uk
In article <na.74aa674eb9.a80620john.no...@orpheusmail.co.uk>, John Nolan <john.no...@orpheusmail.co.uk> wrote:
> Yes, it's me again ! :)
[snip]
> With TopModle2, over the years, I have been struggling with surface > rendering. I think that I have cracked it, but the results seem inconsistent > and the steps complex. The apparent inconsistencies, and the complex steps, > are probably interrelated. You have only to omit, or modify, one vital step > and (obviously) you get a different result. However, if the differences are > subtle, and not obvious and don't "shout" at you, they are not so easy to > spot.
This sort of thing doesn't only apply TopModel!
[snip]
> Some of the "weak links" are where a key flagship application is either > missing, or has problems of difficulty in use. I realise that there is a > chicken and egg situation here. If the RISCOS world is small, developers > jumps ship in order to make a living, so the RISCOS world grows smaller.
I have no intention of jumping ship - BUT my real complaint is the lack of tools to do the job.
Y0ou may talk about Basic or gnu C (gccsdk) and the various interpreted languages (PHP, Python, lua, etc) - but we don't really have tools which I can use to produce - as I would wish - large chunks of software.
I admire people tremendously who can actually write a C program which doesn't fall over. I have been writing programs for over 50 years and I still cannot understand how anyone can use C for anything non-trivial. I only ever look at C to see how things could be done so much more easily in a high-level language - which we don't at the moment have available apart from Ada - which is seen by many as being a bit of a sledge-hammer to crack a nut - less of a problem now though with RISCOS machines having reasonable speed and memory.
> The problem of the idea of "commercial" co-operation is that it is almost an > oxymoron (comparatively new word for me ! :-) ). How do you decide drop the > effort on one of more competing applications in order to make a fewer no. > excellent ? And would the originator want to share it ? Take the example of > ProCAD v. RiscCAD (is it ?). > But I think that this radical, difficult, brave and selfless approach could > be one of the several "shots in the arm" that we RISCOS types all need.
When you only have one pound/dollar/franc and half a dozen kids, they don't get much pocket money each and therefore have a natural tendency to compete. I'm not interested in money (apart from the need to eat and have a roof over my head) or kudos - which is only transitory for most people, I would just like to be able to use high-level languages to produce solutions to (my own at the moment) problems.
In article <na.74aa674eb9.a80620john.no...@orpheusmail.co.uk>, John Nolan <john.no...@orpheusmail.co.uk> wrote:
> Then, coming to the RISCOS world and returning to the "chain". It may be, > that RISCOS could hold it's own, and even make progress, if there were less > commercial competition within its folds and more "commercial" co-operation. > It's not easy to see how this would happen but, if the many "weak links" of > the RISCOS world could be strengthened in a co-operative way, we could all > benefit, include those talented people who make part of their living from > it. > Some of the "weak links" are where a key flagship application is either > missing, or has problems of difficulty in use. I realise that there is a > chicken and egg situation here. If the RISCOS world is small, developers > jumps ship in order to make a living, so the RISCOS world grows smaller.
I understand what you are saying - and there may be times when developers may well be best going to a rival and saying, "I'm developing this in competition with you - can we co-operate?" That may well have worked in one circumstance recently with co-operation allowing merged resources and more sales.
But there is still plenty of 'slack' elsewhere as I mentioned in my talk at the Show on Saturday. There is commercial software that is disappearing and having to be re-invented. We all need to be aware of developers/publishers who may have software that they would be happy to make freely available but who might soon wipe the source from their harddrive if no-one expresses an interest. If such software could be brought back into the fold as freeware - or better still sold at a price that would guarantee continued development - then current developers could avoid having to re-invent the wheel.
-- John Cartmell j...@finnybank.com 0845 006 8822 or 0161 969 9820 Qercus magazine FAX +44 (0)8700-519-527 www.finnybank.com Qercus - the best guide to RISC OS computing
Keith Hopper wrote: > I admire people tremendously who can actually write a C program which > doesn't fall over.
Unfortunately, there aren't many of them, and even people I would describe as "competent developers" have a hard time (and spend too much time) debugging their software.
Someone once said that C should only be used as a "library programming language" because it is rather universally callable from other languages. I tend to agree. However, seeing where RISC OS stands today (where still many lines of code are written in BBC BASIC and Assembler), C would still be an improvement.
> I have been writing programs for over 50 years and I
> still cannot understand how anyone can use C for anything non-trivial. I > only ever look at C to see how things could be done so much more easily in > a high-level language - which we don't at the moment have available apart > from Ada - which is seen by many as being a bit of a sledge-hammer to crack > a nut - less of a problem now though with RISCOS machines having reasonable > speed and memory.
Well, Ada is certainly a rather large hammer, but it is comparatively easy to handle (e.g. you don't usually have to bugger around with makefiles, it automatically detects dependencies and recompiles intelligently).
I always felt that Ada could provide the logical next step for many BBC BASIC developers. I started to write a "BBC BASIC to Ada in easy steps" guide some years ago, and it is surprising how many BBC BASIC structures and commands are easily replaceable by their Ada counterparts. Well, this was another project of mine for the bin...
Unfortunately, Ada on RISC OS is currently let down by two rather major problems: first of all, the available GNAT is incredibly old, and 32bit code generation only works by major hacking, and sometimes shows up major bugs in the compiler/code generation. What is even more important: some aspects of Ada are a bit tedious if you have no or only basic tool support - working with a plain old text editor with a bit of syntax colouring just does not cut it. And finally: GNAT does not work on an IYONIX, even with the help of Aemulor. Which is very unfortunate, because it consumes large amount of memory and would really gain from the fast filing system, faster memory and speedier processor. In the meantime, I use VirtualRPC for the final compile, while using PC-GNAT for the syntax checking part...which does not add to the overall user friendliness ;-)
Maybe some day we will get a recent GNAT based on GCC 4.1.2 which is what the rest of the world uses.
The date being 7 Mar 2007, Steffen Huber <s...@huber-net.de> decided to write:
> Keith Hopper wrote: >> I admire people tremendously who can actually write a C program which >> doesn't fall over.
Eh? I've been involved with development projects using C and have done some programming myself. If a program falls over it's usually because of a logic error which can occur in any language. Yes, you can take advantage of things which C will let you do so you have to know what you're doing. You can get hoist with your own petard if you try to be too clever.
> Unfortunately, there aren't many of them, and even people I would > describe as "competent developers" have a hard time (and spend too > much time) debugging their software.
That's probably because they dived straight into the coding without spending enough time on the design work, or haven't used good structured/modular design methodologies.
> Someone once said that C should only be used as a "library programming > language" because it is rather universally callable from other > languages. I tend to agree. However, seeing where RISC OS stands > today (where still many lines of code are written in BBC BASIC and > Assembler), C would still be an improvement.
Of course. C is a compiled language. Basic is an interpretted language which is quicker to get up and running but a lot slower to execute.
>> I have been writing programs for over 50 years and I still cannot >> understand how anyone can use C for anything non-trivial. I only ever >> look at C to see how things could be done so much more easily in a >> high-level language - which we don't at the moment have available >> apart from Ada - which is seen by many as being a bit of a >> sledge-hammer to crack a nut - less of a problem now though with >> RISCOS machines having reasonable speed and memory.
I fought to have one project written in C rather than Pascal which I absolutely hate. We did have some good programming and inspection standards, and excellent programmers who weren't going write crap code.
Using high level languages is a bit like going by bus instead of driving your own car. You have to go where the bus goes - you have to do what the compiler designer wants you to do. Mind you C isn't exactly a low level language. If you really want to write efficient code you need to use assembler, but you lose portability.
> Well, Ada is certainly a rather large hammer, but it is comparatively > easy to handle (e.g. you don't usually have to bugger around with > makefiles, it automatically detects dependencies and recompiles > intelligently).
> I always felt that Ada could provide the logical next step for many > BBC BASIC developers. I started to write a "BBC BASIC to Ada in easy > steps" guide some years ago, and it is surprising how many BBC BASIC > structures and commands are easily replaceable by their Ada > counterparts. Well, this was another project of mine for the bin...
I've only come across Ada on the defence projects and haven't used it myself (I did use MASCOT and Coral 66).
-- Richard Porter Mail to username ricp at domain minijem.plus.com "You can't have Windows without pains."
In article <557p68F23avl...@mid.individual.net>, Steffen Huber <s...@huber-net.de> wrote:
[Snip]
> Unfortunately, Ada on RISC OS is currently let down by two rather > major problems: first of all, the available GNAT is incredibly old, > and 32bit code generation only works by major hacking, and sometimes > shows up major bugs in the compiler/code generation. What is even > more important: some aspects of Ada are a bit tedious if you have no > or only basic tool support - working with a plain old text editor with > a bit of syntax colouring just does not cut it. And finally: GNAT > does not work on an IYONIX, even with the help of Aemulor. Which is > very unfortunate, because it consumes large amount of memory and would > really gain from the fast filing system, faster memory and speedier > processor. In the meantime, I use VirtualRPC for the final compile, > while using PC-GNAT for the syntax checking part...which does not > add to the overall user friendliness ;-) > Maybe some day we will get a recent GNAT based on GCC 4.1.2 which > is what the rest of the world uses.
If you like Ada, then a visit to Bent Brackes site will interest you ;-).
|
