Left here for archival purposes.

User avatar
By vowstar
#9904
picstart wrote:Well the last working build was
NodeMCU 0.9.5 build 20150123 powered by Lua 5.1.4

this version also appears to work with 512K and 4m Flash

flash is 512K
latest nodemcu prebuild at 115200 baud errors
rl l | l | l b| r b b nn lnn b p lrlrlp n l b n | r b nn l l` nn l nr n b
l` p n r b n | b nn l` nn l nr n bl lp n r lb b n | l l b nn l` nn l
nr n ` r n ` MEM CHECK FAIL!!!
r


Adding float may not be the issue but things have gone awry since float was attempted.

I'm using the nodemcu prebuilds since the windows eclipse IDE by CHERTS does not compile error free for nodemcu-firmware so I can't get the float out of the compile.
The default for float should be to have it selected out of the prebuild's not selected in as it is now.


The "MEM CHECK FAIL!!! " is not floating number bug, It is espressif's 0.9.5 SDK bug, and it is fixed in s espressif's 0.9.6_b1 SDK.
http://bbs.espressif.com/viewtopic.php?f=7&t=205

The new version without "MEM CHECK FAIL!!! " is in NodeMCU github dev096 branch. It is using espressif's 0.9.6_b1 SDK, may buggly.

https://github.com/nodemcu/nodemcu-firmware/tree/dev096
https://github.com/nodemcu/nodemcu-firm ... n?raw=true

You could clone and compile it, and if you want, you can disable the floating point by your self. If your can't compile it, could you copy the error log and give NodeMCU an issue?
User avatar
By picstart
#9907 Thanks vowstar espressif's 0.9.6_b1 SDK fixed the MEM CHECK FAIL!!! issue.
I was compiling with the windows eclipse IDE from CERTS . For some of my code it would compile error free but would at run time bomb with MEM CHECK FAIL!!!
I didn't suspect it was 0.95 SDK issue...anyway the time absorbed looking into my code wasn't a total waste.
"You are smart if you learn from your own errors but wise if you learn from the errors of others"
Yes, espressif you may be making me wiser as I learn from errors in the 9.5 SDK
User avatar
By Fr4gg0r
#9975
Theo wrote:
jankop wrote:I'm against it. If not rise significantly memory requirements of programs, so it is clearly beneficial. I even edited the first program for FP.

> =(600-600%70)/70
8
>

:D


This does not work for negative numbers !


And it is annoying.

Is there no rounding function available?