chilipeppr
commented
6 years ago I think your code change looks reasonable and if you've tested it on a couple of jobs then you're probably good to go and a pull request would be welcomed.
If you look at the TinyG buffer, I think I remove the \n in there on a feed hold, resume, or wipe. The thing about \n is that SPJS is very much a line-centric approach, like if you send in a bunch of text with \n's in the middle, it splits and turns those into lines that are sent in that order to the serial device. Generally CNC firmware is line-centric, but then you have the non-line centric stuff like you point out such as !, ~, and %. So, it does make sense to drop the \n in those cases.
On Mon, Feb 5, 2018 at 10:15 AM, setaperlacloche notifications@github.com wrote:
(Context SPJS 1.95, Grbl 1.1f on Arduino Nano)
When I run a GCode file and do a lot of hold/resume commands, I get errors (mainly 25 et 33).
In order to guess the reason, I put a logical probe on RX/TX on my Arduino. And discovered that '!' (Feed Hold command) and '~' (Cycle Start / Resume) are followed by '\n'. In fact all realtime commands (including jog, fro, Spindle Speed Overrides,..) except '?' are followed by a '\n'. This is a violation of the protocole about realtime commands (from the documentation) :
- Is a single character that may be sent to Grbl at any time.
- Does not require a line feed or carriage return after them.
- Is not considered a part of the streaming protocol.
When a "!/n" is sent, SPJS considers this command skip the internal serial port buffering (serial.go:412), but on Grbl side, char '!' skip the buffer but not the '\n' ==> At this point bytes counting in buffer are unsynchronize between SPJS and Grbl. Grbl hold one more character than SPJS thinking.
The extra '\n' is added in bufferflow_grbl.go:269 So I replaced the line that detects '?' as the only realtime command (bufferflow_grbl.go:260) from :
} else if item == "?" {to :
} else if b.SeeIfSpecificCommandsShouldSkipBuffer(item) {With this modification, it seems that buffer overflow is away.
What do you think ?
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/chilipeppr/serial-port-json-server/issues/42, or mute the thread https://github.com/notifications/unsubscribe-auth/AHidbYzSpTVdDpTaIYOQ3I6hdg4EAvW3ks5tR0UwgaJpZM4R536r .
setaperlacloche
(Context SPJS 1.95, Grbl 1.1f on Arduino Nano)
When I run a GCode file and do a lot of hold/resume commands, I get errors (mainly 25 et 33).
In order to guess the reason, I put a logical probe on RX/TX on my Arduino. And discovered that '!' (Feed Hold command) and '~' (Cycle Start / Resume) are followed by '\n'. In fact all realtime commands (including jog, fro, Spindle Speed Overrides,..) except '?' are followed by a '\n'. This is a violation of the protocole about realtime commands (from the documentation) :
When a "!/n" is sent, SPJS considers this command skip the internal serial port buffering (serial.go:412), but on Grbl side, char '!' skip the buffer but not the '\n' ==> At this point bytes counting in buffer are unsynchronize between SPJS and Grbl. Grbl hold one more character than SPJS thinking.
The extra '\n' is added in bufferflow_grbl.go:269 So I replaced the line that detects '?' as the only realtime command (bufferflow_grbl.go:260) from :
What do you think ?