infobomb
Member
In the arduino compatibility notes it says String is not implemented. In the 0.0.9 notes is say it is. I'm trying to port some code that passes:
void bar(String foo)
as a function arg and it gave me strange errors until I changed void to unsigned int. Then I get another strange error:
error: 'String' was not declared in this scope In function 'void setup()':
My code works fine with arduino 0018. I'm starting to think String isn't implemented in 0.0.9.
Any ideas?
infobomb
Member
More data:
I built a bare sketch with only this (I wasn't sure String was in the core):
#include <string.h>
String test_input = "0012224826351340262441240122631320562191191OAT00090TRE-0061FLP0001020481378139214061421143514503583533633743843951103D2";
And got this:
error: 'String' does not name a type
UPDATE: Tried 0.0.10Beta - no change.
The docs at http://static.leaflabs.com/pub/leaflabs/maple-docs/0.0.10beta/lang/cpp/string.html#lang-string appear incorrect in suggesting that String exists. No problem - I'll write a wrapper until it works.
robodude666
Member
From what I've gathered, the String class is mentioned nowhere in libmaple; arduino compatibility page, or notes. If you look at libmaple, there is no String class file either.
If you must Arduino's string class, you can use their string class =].
P.S. string.h includes the standard cstring functions.
mbolivar
LeafLabs
infobomb,
Oops, you're right, those docs are in error. Sorry for the mistake. I don't remember how that ended up in there; probably I meant to include String by 0.0.9. After considering the matter, however, I'm not really sure that the String class is a good idea on a board with such a limited amount of memory, so the plans to include it got scrapped.
If you watch the Arduino developers mailing list, you may have seen that they've been having tons of trouble getting the semantics right for String; error handling in particular seems like it's been a real nuisance.
If it's any consolation: since Maple Native has that nice 1 MB SRAM chip, there are plans to include support for dynamic memory allocation in its library support (although the first IDE release will likely not include this). Users of that board will thus likely be able to just use std::string.
infobomb
Member
That's kind of what I thought (on both counts). I used String out of laziness on the arduino but I did note the bloat. I'll recode for both versions and save some bytes - no problem. I'm only doing sub string stuff - no matching.
The Maple's a nice little board, BTW. I'd love to have both 5v and 3.3v Vcc on the board someday as the primitive LCDs I want to power won't run on 3.3.
DMA on a one chip board. Wow. I remember my old Apple days when we were so excited to get it on a Mac... ;-)
Thanks for the response folks!
mbolivar
LeafLabs
Glad to hear that this wasn't a deal-breaker for you, and again I'm sorry for the mistake. I've fixed the uploaded 0.0.10beta docs:
http://static.leaflabs.com/pub/leaflabs/maple-docs/0.0.10beta/lang/cpp/string.html
Fix was pushed in c0721a7:
https://github.com/leaflabs/libmaple/commit/c0721a714484446b0b4a30c057d8b01df14b088f
infobomb
Member
I recoded my sketch and built my own (very limited) atoi and eliminated the need for one other String method. The LCD now works with my Maple... for a while. At some point (a few minutes of 115,200 baud data, 120 bytes, 64 hz) the LCD display goes nuts. I'm suspecting something bogus is getting sent to the LCD.
I then took the non-String code back to my Arduino (Seeeduino 2.2.1) and it compiled in 4k instead of 8k. You folks were right, the String class was big (by AVR standards). Interestingly, the code works here with no LCD insanity creeping in. Several 1 hour runs with no faults. On both Maple and Arduino tests all components except the main boards were the same (LCD, MAX233A, USB to Serial for input data to RX1). The only possible difference is that I "borrowed" 5vcc from the seeeduino to power the MAX233A and the LCD and shared grounds with the Maple. I didn't put a scope on it but I sort of doubt this is the problem.
SO: Is there a way to instrument the LCD Library on Maple IDE so I can delve a bit deeper into the mystery. I'd like to figure out what's causing the LCD issues for the community. I have info from the Arduino site on how to make a contributed library. I'll try taking the LiquidCrystal LIB for Maple and doing a test version. If there are any cool tools that help with this please let me know!
Thanks - I hope to have some info in a few days.
mbolivar
LeafLabs
infobomb,
Sorry to hear you're having more problems. Thanks for your offer to help!
The LiquidCrystal library is a port of the Arduino version, so there's a large shared set of code between them. I fairly recently made some changes which shave time off of LiquidCrystal::pulseEnable() that may be the culprit, although to my (admittedly very limited) knowledge, these are OK according to spec. More information is available in the pulseEnable() source:
https://github.com/leaflabs/libmaple/blob/master/libraries/LiquidCrystal/LiquidCrystal.cpp#L300
The best way I know of to instrument code running on the Maple is with JTAG. When used in concert with OpenOCD, this buys you the ability to attach arm-none-eabi-gdb to the code running on the Maple.
There's a HOWTO up on the wiki which contains information on making a JTAG adapter for the Maple (and on OpenOCD, etc.):
http://wiki.leaflabs.com/index.php?title=Maple_JTAG_How_To
Thanks to a collaboration of forum users, we have a limited number of JTAG adapter kits for sale; if you're interested in one, please email us at info@leaflabs.com. (We'll be doing larger runs of these in the future).
You must log in to post.