Freeze when opening certain files about p2-vscode-extensions HOT 7 CLOSED

Hmm... seeing spin code in .spin2 files and spin2 in .spin files. They have a lot of red because of incorrect language in the given file, BUT I'm not seeing any lockups.

hexagon_sdda.spin2 (Spin V2 language) contains spin (v1) object constant uses: Ex:

line(61) erc:=\fat.FATEngineStart(plat#SD_DO,plat#SD_CLK,plat#SD_DI,plat#SD_SELECT,false)

hexagon.spin (Spin V1 language) contains spin2 method uses: Ex:

line(333):   wrpin(plat#CONTROLACC_BASEPIN+4+(3<<6),P_NORMAL|P_LOW_150K|P_HIGH_FLOAT)
line(334):  pinl(plat#CONTROLACC_BASEPIN+4+(3<<6))

platform.spin (Spin V1 language) contains non spin(or spin2) object declaration: Ex:

line(48): OBJ video = "p2videodrv.spin2"

This should be a colon vs. an "=" separator.

Wow, and the Syntax parser is not coloring the platform.spin file. This I'm looking into...
Um, NOPE, never mind. This appears to be due to the syntax parser being unable to handle the undeclared first section of CON. This is a known bug.

from p2-vscode-extensions.

I have no plans to support mixed p1 and p2 syntax in the same file. So I'm not sure I can do anything about these.

What do you think should be fixed if there's no lockup?

I am adding error reporting for object declarations that I can't parse!

from p2-vscode-extensions.

The = style object declaration is a flexspin feature. It means something special (reference class but don't instantiate an instance - you can use the subscript operator to then access an instance through a pointer - also good if you just need to reference constants)

from p2-vscode-extensions.

sdspi_with_audio.spin2 (Spin v2 language) missing org directives at start of DAT (required to enter pasm parser.) ex:

dat 'do_init     'RJA disabled smartpins
do_init

                getbyte pinDO,          arg1_in,    #0
                getbyte pinCLK,         arg1_in,    #1

The parser fails to handle this correctly as it is not data declarations so it flags lots of random bits.

Also found a missing comment "'" at:

con  smartpin_constants

so it flagged smartpin_constants as a constant expression missing the '=' assignment part.

from p2-vscode-extensions.

FOUND: VJET_p2beta_displaylist.spin2 (Spin V2 language) 1 line is causing the parser crash:

PUB line_polygon(polyptr,vcount,width,colors) | i,px,py

  px := long[polyptr][(vcount<<1)-2]
  py := long[polyptr][(vcount<<1)-1]
  repeat i from 0 to vcount-1
     line(px,py,px := long[polyptr][(i<<1)],py := long[polyptr][(i<<1)+1],width,colors)

The last line causes a crash. It shouldn't happen.

Hrmf... not a crash per see, my new parser was mis-tokenizing the statement then couldn't find the mashed parts. so generating errors with bad char offsets.

So, let me understand this code.

I think you are implying this:

PUB line_polygonSimple(polyptr,vcount,width,colors) | i, px, py, px2, py2

  px := long[polyptr][(vcount<<1)-2]
  py := long[polyptr][(vcount<<1)-1]
  repeat i from 0 to vcount-1
    px2 := long[polyptr][(i<<1)]
    py2 := long[polyptr][(i<<1)+1]
    line(px,py,px2,py2,width,colors) 

...with you intending to save from allocating two more variables but counting on the compiler's order of operations to keep the old and new px and py values correct.

Am I right?

from p2-vscode-extensions.

yes, that's roughly how that works. It wants to draw lines between points, so each point needs to be used twice. The previous point is stored in px/py, which is updated in the same statement where it just got used.

from p2-vscode-extensions.

Appears to be fixed in current version

from p2-vscode-extensions.