I really *really* like the look of "FlowCode"! :-D

[IanB]

True, but math isn't something that really goes anywhere. You can see you need some math but beyond add/sub/mul/div students find it hard to care much about it.  It's needed for problems but it really doesn't do anything (except maybe PHD level math on cutting edge science).

Wow, that really fails to understand what mathematics is all about. Such a view is probably an outcome of the failed teaching of mathematics in schools.

Mathematics is about conceptual and logical thinking, about seeing patterns and connections in things, about using this to build up from simpler ideas to more complex ideas while keeping things straight in your mind. Mathematics is not merely about sums and algebra and solving problems, although such basic mental calisthenics are a natural part of a learning program.

It's really sad the the teaching of mathematics in schools is so often reduced to the routine teaching of simple rules and procedures. It does a great disservice to the children exposed to it by taking a wonderfully creative subject and reducing it to tedious boredom.

[Psi]

True, but math isn't something that really goes anywhere. You can see you need some math but beyond add/sub/mul/div students find it hard to care much about it.  It's needed for problems but it really doesn't do anything (except maybe PHD level math on cutting edge science).

Wow, that really fails to understand what mathematics is all about. Such a view is probably an outcome of the failed teaching of mathematics in schools.

I was describing it from the point of view of the student.
Doing school math for no other reason than getting an answer doesn't lead to anything. Doing programming leads to an output, something semi-tangible that you can see a use for.

It was a comment on BoredAtWork's statement that  "most people suck at simple math after they finished school"
They suck because doing math for no reason other than doing math has no value to them. For young people to see value in something it needs to have an output other than the learning itself.

And as you say. it should be about teaching conceptual and logical thinking.

[free_electron]

but then SW pros don't need comments because the source is self commenting, right.

Real programmers write hex straight into mask-rom. all this namby pamby stuff with high level languages , compilers , flash , retry , breakpoints debuggers is for beginners. Real programmers know the whole instruction set in its hexadecimal form ( opcodes are for 'intermediate programmers' ) and write flawless, elegant code that you can directly put into mask rom. It will work first time right. Anybody that does it different is a 'wanna-be' (that includes myself ).

@steve_w : I sometimes write 'critical' code and there is a whole bunch of rules.. no dynamic memory allocation, no function pointers . We can use 'C' but a whole bunch of constructs are 'illegal'. We have a pre-processor (a kind of linter) that goes over the code and blocks compilation if we do anything deemed 'potentially unsafe'. code is profiled the snot out of it. every pathway is walked to make sure the code can complete its task, barring severe hardware problems, in the alotted amount of time.

I don't know about flowcode and i'm not going to comment on it but , from what i have seen : it looks like a graphical wrapper around standard 'c'.
In essence you drop down a block , define in and out parameters , doubleclick it and fill in the code. This i kind of like. In a modern day and age it is stupid that you have to define a function prototype in the .h file , then type exaclty the same thing again in another file. If you need to modify the parameters , you need to go back in the .h and tweak the prototype .. there has to be an easier way. These are the boring ridiculous 'typing jobs' that can easily be replaced by a smart IDE. i think that is what flowcode attempts to do. Take the tedious, boring, time consuming , error prone , easily overlooked things away so you can focus on writing the functional part of the code.

Why do we even need things like forward declaration anymore ? just make the compiler multipass. first pass collects all the functions and compiles them. whenever there is a call to an , as yet unknown , function : leave a marker behind. once the first pass is done you now have a map. go back and replace the markers with the final function address... this is a 'mechanical' job that can perfectly be done by a compiler.

[FreeThinker]

but then SW pros don't need comments because the source is self commenting, right.

Real programmers write hex straight into mask-rom. all this namby pamby stuff with high level languages , compilers , flash , retry , breakpoints debuggers is for beginners. Real programmers know the whole instruction set in its hexadecimal form ( opcodes are for 'intermediate programmers' ) and write flawless, elegant code that you can directly put into mask rom. It will work first time right. Anybody that does it different is a 'wanna-be' (that includes myself ).

@steve_w : I sometimes write 'critical' code and there is a whole bunch of rules.. no dynamic memory allocation, no function pointers . We can use 'C' but a whole bunch of constructs are 'illegal'. We have a pre-processor (a kind of linter) that goes over the code and blocks compilation if we do anything deemed 'potentially unsafe'. code is profiled the snot out of it. every pathway is walked to make sure the code can complete its task, barring severe hardware problems, in the alotted amount of time.

I don't know about flowcode and i'm not going to comment on it but , from what i have seen : it looks like a graphical wrapper around standard 'c'.
In essence you drop down a block , define in and out parameters , doubleclick it and fill in the code. This i kind of like. In a modern day and age it is stupid that you have to define a function prototype in the .h file , then type exaclty the same thing again in another file. If you need to modify the parameters , you need to go back in the .h and tweak the prototype .. there has to be an easier way. These are the boring ridiculous 'typing jobs' that can easily be replaced by a smart IDE. i think that is what flowcode attempts to do. Take the tedious, boring, time consuming , error prone , easily overlooked things away so you can focus on writing the functional part of the code.

Why do we even need things like forward declaration anymore ? just make the compiler multipass. first pass collects all the functions and compiles them. whenever there is a call to an , as yet unknown , function : leave a marker behind. once the first pass is done you now have a map. go back and replace the markers with the final function address... this is a 'mechanical' job that can perfectly be done by a compiler.

After playing with it for a day or so I can confirm that it is indeed simply a wrapper around a c complier. You can add code and (I think) even alter the predefines macros so you can drill down quite deep to the core. You can down load a crippleware version from the website to play with.

[Bored@Work]

Real programmers write hex straight into mask-rom.

Bullshit. Real programmers write binary directly into mask ROMs by bending the path of the photons in the photolitographic phase of ROM creation with bare hands. Non of that modern hex stuff. Bah.

[Mechatrommer]

but then SW pros don't need comments because the source is self commenting, right.

SW pros do need comments, alot of it. code monkeys dont.

Real programmers write hex straight into mask-rom. all this namby pamby stuff with high level languages , compilers , flash , retry , breakpoints debuggers is for beginners. Real programmers know the whole instruction set in its hexadecimal form ( opcodes are for 'intermediate programmers' ) and write flawless, elegant code that you can directly put into mask rom. It will work first time right. Anybody that does it different is a 'wanna-be' (that includes myself ).

with your postulate, i believe everybody is wannabe including the "Real programmers write hex straight into mask-rom". the only one i know that doesnt make mistake and "work first time right" is God. pls name "human" if you know any. i'm positive it wont be "Bill" or "Steve"

[SeanB]

Real programmers write hex straight into mask-rom.

Bullshit. Real programmers write binary directly into mask ROMs by bending the path of the photons in the photolitographic phase of ROM creation with bare hands. Non of that modern hex stuff. Bah.

You write the bits in by plugging a fusible link prom into a breadboard, connecting a power supply and blow bits using wire links to set up the address and data then briefly powering the prom to blow the fuses. Followed by reading with a voltmeter after setting the power supply back down to 5V from the programming voltage of 6V. Other option was wiring a core memory with sense wiring to make a read only core.

[Mechatrommer]

real programmers only need a punch and hardpaper card, no electrons and nics and inductism is involved.

[A Hellene]

Real programmers write hex straight into mask-rom. all this namby pamby stuff with high level languages , compilers , flash , retry , breakpoints debuggers is for beginners. Real programmers know the whole instruction set in its hexadecimal form ( opcodes are for 'intermediate programmers' ) and write flawless, elegant code that you can directly put into mask rom. It will work first time right. Anybody that does it different is a 'wanna-be' (that includes myself ).

with your postulate, i believe everybody is wannabe including the "Real programmers write hex straight into mask-rom". the only one i know that doesnt make mistake and "work first time right" is God. pls name "human" if you know any. i'm positive it wont be "Bill" or "Steve"

Well, without claiming any titles --any titles at all-- I have recently done that!

Specifically, I patched the DS1000EUpdate.RGL v2.06 (the Rigol firmware update binary file), replacing a code chunk with a custom one, inspired by Krater's Homebrew work on the firmware v2.05.

What I did, was to:
1. Spot the chunk of code to be replaced in the disassembly listing of the image loaded into the SDRAM system space,
2. Write in assembly the piece of code I needed to replace the original one with,
3. Convert the instructions to op-codes,
4. Manually calculate the source and the target function and buffer offsets,
5. Write down the corrected op-codes, without invoking any Blackfin assemblers or linkers,
6. Calculate the offsets of the new chunk of code within the executable image file (the "DS1000EUpdate.RGL") and replace the original code chunk,
7. Confirm in disassembler that the new image works as expected,
8. Load the new firmware and, finally,
9. Successfully test it!

How much time did all this take for me to do it? Almost two working days (a day and a half, actually)!

Why I did I choose that path? Because of the minimal time it required from me to do it, by only using the Blackfin documentation (the ADSP Instruction Set, the Hardware and the Programming references and the Boot-ROM Engineer Note), without having to get myself to learn the Analog Devices VisualDSP platform.

So, yes, there certainly are many ways for a job to be done!


-George

[Mechatrommer]

thats called hacking, not programming. and worst, copycatting with luck
or are you claiming you are better than rigol programmer who built the "buggy FW" from scratch?

[A Hellene]

You can call it anyway you like; I've already stated that I do not claim any titles for that!

What I claim is that the modification described above was built from scratch, it was applied directly to the op-code level with the aid of a calculator (without using any fancy GUIs or any other programming band-aids), it was really fast to be done, and it worked right from the first time.

Oh! And that I am human!


-George

[Kremmen]

but then SW pros don't need comments because the source is self commenting, right.

Real programmers write hex straight into mask-rom. all this namby pamby stuff with high level languages , compilers , flash , retry , breakpoints debuggers is for beginners. Real programmers know the whole instruction set in its hexadecimal form ( opcodes are for 'intermediate programmers' ) and write flawless, elegant code that you can directly put into mask rom. It will work first time right. Anybody that does it different is a 'wanna-be' (that includes myself ).

Been there, done it. Used to be i knew the 8085 opcodes in hex almost all by heart. Needed them too as the only ISIS-2 development systems were expensive, big and definitely did not leave the office. All field fixes were done directly to 27xxx UVEEPROMS using a manual programmer. I don't miss those days.

but then SW pros don't need comments because the source is self commenting, right.

SW pros do need comments, alot of it. code monkeys dont.

That was sarcasm. I know for the amerikans the word "irony" means something with lots of iron in it. You come from elsewhere so you should do better  . As a rule, I have my code monkeys read "Code Complete", in my opinion one of the best books about practical level code construction. After they assimilate the book, they are not monkeys any more. They also know how to annotate the code in a meaningful way.

While on the topic of Real Programmers: The Story of Mel http://www.pbm.com/~lindahl/mel.html.

[free_electron]

yay ! an isis-2 ! the blue one or the more advanced white one with built in terminal ?
i had one of those with the S100 blade with one of those rainbow flatcables ( twisted pair version ) and a real ICE-51 box at the other end.
Later we also had those for the i960 ... now THAT was a processor... when everybody was still mumbling about 'risc is the future' that thing already ate anything our there...

Isis.. and iRMX .. those were the days. Compiling PL/M code. iRMX is one of the most bulletproof operating systems out there.

Best 'error trap' i have ever seen on an iRMX system:

'Beep' .... in the console status line appears a little text : Parity error in RAM ,Pages 800 to 816 ( or something along those lines ). Contents redirected. Autocontine.

that was it. just a little message that part of the computers live RAM had died , it had relocated these pages , reloaded the data and went on about its business.

The bloody damn thing just continued running as if nothing had happened ! it just had one of its ram chips on the motherboard fail.. completely ( the ram chip was dead ) the iRMX kernel detected this , located what was stored there , retrieved it from disk , allocated memory somewhere else for this , changed the runtime on the fly to use the updated addresses and marked these ram blocks int he memory manager as 'bad' so it would not go there. The whole machine kept on running for weeks without a single failure... then it was time for a scheduled maintenance. we took it down . located the bad memory ( They used 42256 chips. This was an old 80386 pc board from Radisys). we pulled the chips out of the sockets , put some new ones in , booted in manager mode , ran the memory test to verify the ram was now good and iRMX reenabled those pages... and happily thrundled off doing its business..

Any operating system that can pull something like that off is a real operating system. Here you have this catastrophical hardware failure and it just relocates code , puts a little message on the console , marks the offending hardware as 'do not use' and keeps on running ...
I'd like to see you pull a stunt like that on anything else ...

[free_electron]

?? what ego battle ? i was talking about iRMX and the isis systems.

Has nothing to do with what i did .. i couldn't begin to make such a thing ...
There's plenty of people that have done waaaaay more interesting things in life than the little bit of dabbling i have done ... Like Gary Kildall when he made the PL/M language and wrote CP/M....

[steve_w]

Free_electron

The point I was making was that code generators dont allow the inspection that you describes, you dont know how the code is implemented.  If you write it yourself, then it can be peer reviewed and walkthroughed and checked for conformance to standards.  If you have a flow chart and an executable and a code generator what do you inspect?

regards

SW

[EEVblog]

After playing with it for a day or so I can confirm that it is indeed simply a wrapper around a c complier. You can add code and (I think) even alter the predefines macros so you can drill down quite deep to the core. You can down load a crippleware version from the website to play with.

I've been told by a user that this is indeed the case. You can use inline C any time you want.

Dave.

[A Hellene]

... and so the ego battle has started; "I've done something to upstage YOU". Bah! ;-)

Well, in the case of that "ego-battle" statement being an aftereffect of my messages, I just stated that no fancy software would be really needed if someone really knew what they were doing.

Now, my previous message was a tease to our co-EEVBlogger who asked if there was any known human being to be programming straight in hex, since he wrote that "only God acts without making any mistakes." I also stressed that I did not claim any titles for what I did, nor am I the Tetragrammaton (= YHWH: "he who is") because I am unreligious (meaning that I am indifferent to religions and I do not recognise any religious dogmas as authorities in my life, as we had already discussed that in the past).


-George

[free_electron]

Free_electron

The point I was making was that code generators dont allow the inspection that you describes, you dont know how the code is implemented.  If you write it yourself, then it can be peer reviewed and walkthroughed and checked for conformance to standards.  If you have a flow chart and an executable and a code generator what do you inspect?

regards

SW

as i understand it this is only a wrapper around a bunch of c libraries and your code ( which is also c). if they provided the source to the libraries ... you are set.

[HackedFridgeMagnet]

I am not one for Nationalistic flame wars but I LOL when I read.

I know for the amerikans the word "irony" means something with lots of iron in it.

 

As a rule, I have my code monkeys read "Code Complete"

Great book. Handed to me when I was turned into a code monkey, that and the directive that I had to learn to touch type, so I could type out all those long method names.

[Mechatrommer]

Now, my previous message was a tease to our co-EEVBlogger who asked if there was any known human being to be programming straight in hex

i believe we've been talking different terminology here. where for me, programming is... 'at some level of intelligence' code execution/creation. not 1+1=yada whatever it is. edit and by 'at some level of intelligence' i mean something that cannot be done in 2-3 days time, no matter how correctly and hardly you do it. YMMV.

I dare say: he/she/it, if existing,  screwed up really bad when creating the human species

its by design! whether you want to accept it or not, your choice thats the way how it works, god's choice.

btw a food for thought. if you see one ant raping or killing the other ant what will you do? flat it on the floor with your thumb? excessively off-topic

[Mechatrommer]

btw. the hardest part that recently came across in my life is 'programming silicons and electrons' the way we want them to be. i never faced such difficulty in software domain.

[Mechatrommer]

i tried reply #57, but as usual its unstopable. so.... just following the 'flow' or the 'code' and i didnt take any offense, i'm... 'neutral' and its not night around here, infact 30 minutes to noon

[Mechatrommer]

maybe later. i've no intention going SW right now. i'm stucked at the analog may give it a try next time. thanks for sharing the info, it maybe the "future"!

[Mechatrommer]

and more time with PHYSICAL hardware

as i get older i prefer the easier. i partly understand why people prefer the premade chip instead of discrete transistors. maybe this is one time only i delve so deep into the analog for an exercise. after this maybe i'll just get back to arduino and code somthing fun. enjoy life right? before we embrace 'nonexistency'