I have done some embedded work, but not with such small ram limitation. The first thing you need to do is to see what version of gcc [clang] (and libs) you may be dead-end at based on perhaps drivers, vendor restriction, etc. Maybe your not restricted at all, you can push a full new linux kernel and trimmings to it. If you can get C++17 or 14 then things are looking great for the solution of C++ and all the new toys it has! Also benefit - people don't steal your script code (but you can always go to a native compiler for python maybe). You would never write cgi in C unless you are dead-ended on gcc support prior to C++11(ish). I think I got wt (web toolkit) down to a small foot print, and it would make an amazing web platform on that foot print. regex lib was big trouble , as I was targeting < 200MB ram, but that was because other stuff had to run. If you need a small regex lib, I can give info to you. As for saving state, depends on amount of data, if you are taking about session state, etc, some embedded devices come with libs for that, maybe some using sharedmem. if state does need to stay over reboot, then want to avoid burning flash cycles on DB (of course). I am pretty sure you will a decent chance of finding a key-pair ram storage ability provided (or available lib), unless its a very odd platform. If you need across boot persistence, then you have to look at a some flushing and startup loading. Interpreted languages give you garbage collection, reflection, dynamic interp. execution, say, over C++(old). With C++(new) new memory management options, , memory leaks, etc are pretty much way of the past., reflection can be tackled pretty well if you have control of code (meta data reflection libs), but if you need dynamic script execution, then of course c++ is out, unless you have much more memory and can bring in the C++ interp. to run dynamic C++ scripting (i.e. .Cling). On such a small footprint you almost certainly can't run [Cling] C++ in its interpreted language format. -tl On Tue, Jan 15, 2019 at 10:53 PM William Park via talk <talk@gtalug.org> wrote:
On Tue, Jan 15, 2019 at 07:49:15PM -0500, Kevin Cozens via talk wrote:
On 2019-01-14 12:35 a.m., William Park via talk wrote:
It so happens that I'm looking for interpretor suitable for embedded applications. I read up on "Lua". Maybe there are other options?
Without knowing your intended use case(s) it is hard to know what language(s) would be considered suitable. Other options are implementations of interpreted C, Forth, or dare I say, BASIC.
My working environment is 256MB storage, 256MB ram, F2FS filesystem, ARM cpu, stripped down Linux kernel, and Busybox. Most things are written in C. But, comments and requests from customers, nowdays, are more "web" direction. So, if we write web apps, I'm wondering whether we shoud write all those CGIs in C or some interpreted language. It doesn't have to be that fast, as long as it's not too slow. :-)
Main feature I need is ability to save "state" of some data structure, say variables, array, or dictionary, without having to parse/reparse when writing/reading from filesystem. Python can do that. I can do that in C too. My last choice would be SQLite, though, it has its advantages. -- William Park <opengeometry@yahoo.ca> --- Talk Mailing List talk@gtalug.org https://gtalug.org/mailman/listinfo/talk