Yep good catch Tim, the StrongHelp manuals are a neat way to quickly look up this or that SWI when you need it.timrowledge wrote:Don't forget the StrongHelp manuals; they're so easy scan through and search that a lot of the time they are much more useful than the PRMs.
For SWIs in particular it's worth getting the os335 manual and the oslib one.
See http://www.riscos.info/downloads/stronghelp/manuals/ for a loooong list.
Once you learn the api in BBC BASIC V it does not change for other languages. Though yes it would be atractive to add tutorials for other languages to attract new developers.dr_d_gee wrote:As someone used to other platforms I'd say RISC OS has a lot of technical reference documentation, but little tutorial information; what there is is largely targeted at BASIC and possibly leaves too much out.
I had started work on a series of tutorials for new RISC OS programmers. My goal is to give pretty much equilivent tutorials for:What's needed (for C/C++ anyway, IMHO) is a RISC OS equivalent of Charles Petzold's classic "Programming Windows", which focusses on things you want to *do* and shows you how to do them.
What is wrong with !WinED + !Zap (or simular) + GCC (or Acorn C/C++)? These tools provide a more complete, easier to use, and equaly modern development enviroment for a C programmer than those available for Linux. And because of the standard Drag and Drop paridigm for managing files, and copying text or other media between applications it is quicker to have the three applications than to use an IDE.The other problem with attracting developers to RISC OS is, quite frankly, that the development tools themselves are rather outdated. There isn't an IDE in the way in which such a thing is now understood. Compare the RISC OS tools with QtCreator for example--code completion, integrated GUI designer, integrated (extensive) documentation, both reference and tutorial--and it's free (provided you're happy to distribute programs under the GPL--otherwise a commercial license is *very* expensive).
We have a better development enviroment if your goal is productivity see above. Do not improve on perfection you will just slow yourself down.Even other 'minority' operating systems such as Syllable and Haiku have 'modern' IDEs (even if few other applications). So I suspect attracting and keeping developers--especially those with no deep commitment to the OS--may prove difficult.
And Gnome uses GTK, and a few other dynamic librarys that add an extra layer of abstraction on top of the WM itself ontop of X11. If you want to use C++ no one is stopping you from writting a __packed__ class that for each of the 4 structures and adding the methods to manipulate the the class data elements and call the SWIs.dr_d_gee wrote:The other issue is that apart from using an obscure language like Charm (unknown apart from RISC OS) developing a GUI (in a readable fashion) involves using C rather than C++. Why was Vala developed? To provide an alternative to C#/Mono for developing programs for the GNOME desktop (on Linux). Why was C#/Mono being used? Because the native GNOME language was C, not C++, and C# is a lot more productive.
If you have the GUI front end open for your compiler, or make (or another project manager that you prefer) then it is just a matter of dropping the contents of the build file (eg make file etc) on the corosponding IconBar Icon. This is a lot easier and faster than the methods used in the cumbersome slow wasteful IDEs on Linux + x11 + WM.While !Zap and !StrongEd are good editors, they are not really as capable as SCITE (an editor for Linux--and other platforms--that can allow you to compile from within the editor, and doesn't need the compiler to support "throwback" to provide a list of error messages, clicking on one of which takes you to the relevant line).
!Zap has no trouble with getting file types correct, if it is /c or in the directory c. then it is C, /o or directory o. an object, /s ordirectory s. it is assembly etc. And files that have an alocated type (BASIC, Obey, etc) there is no need for filename extensions.And one of the useful things about file extensions is that the editor can tell what language it is supposed to be, and adjust accordingly...
Realy, then I assume that you have never used !WinEd?Even then, for creating GUI applications *none* of these tools is as easy to use as something like QtCreator. Or Lazarus. Or...
Agreed, I've worked in Delphi in the past and use Visual Studio currently and the amount of convenience these supply shouldn't be underestimated. In VS, for example, autocompletion, code snippets and pop-up context sensitive help as you type (intellisense) are a real boon. Something like that on RISC OS would make the issue of SWI documentation moot - as you typed you'd get the info you needed (and more extensive help would be a small hop away).dr_d_gee wrote: Even then, for creating GUI applications *none* of these tools is as easy to use as something like QtCreator. Or Lazarus. Or...
Thank you for that. That is getting darned close to being a full blown IDE, and since so many developers are stuck on that point may help attract new developers.
Thanks very much for that Microbits, I'll give it a look.
People are stuck on that point for good reason David, anything that makes development/debugging and testing easier is a good thing (IMHO).DavidS wrote:Thank you for that. That is getting darned close to being a full blown IDE, and since so many developers are stuck on that point may help attract new developers.
I agree with that. That is part of why I enjoy developng on RISC OS.AMcS wrote:People are stuck on that point for good reason David, anything that makes development/debugging and testing easier is a good thing (IMHO).
Yes I understand that many of the RISC OS newcomers will be accustomed to the Linux adn Windoze syle IDEs, I have worked on both systems, using some of these IDEs for years at a time. While they are the best available option on those platforms, they slow you down as compared to the way we do things on RISC OS.To develop RISC OS we need more developers - some will come from "outside" the historical cohort so anything that "eases" their path into RISC OS so much the better. For existing RISC OS developers it would make the task of developing (particular WIMP based apps) less fraught and allow for increased productivity without placing unreasonable demands on them or their time.
I have not looked at it very much yet, though it does apear to be a partial solution. I think for a complete solution someone would have to put in a week and create a coplete IDE, and then however long to extend it to the needs of RISC OS Newcomers.As I said to Microbitsuk I'll give Sourcery a look and see if it is a solution or at least a part of the solution desired.
I'd agree it is a partial solution, there's a lot of "fiddling" about with settings to do more complex builds and from the documentation it does seem that Sourcery would much simplify that (good !). For a competent Linux coder (say) it would flatten out some of the issues so they could concentrate on coding without wondering with switches to set in Norcroft or how to create Obey or Run files.DavidS wrote:I have not looked at it very much yet, though it does apear to be a partial solution. I think for a complete solution someone would have to put in a week and create a coplete IDE, and then however long to extend it to the needs of RISC OS Newcomers.