search

3Dickulus / FragM

Derived from https://github.com/Syntopia/Fragmentarium/
GNU General Public License v3.0
344 stars 30 forks source link

don't jump to error line if cursor is nearby #156

Closed claudeha closed 3 years ago

claudeha commented 3 years ago

Describe the bug When editing, the error location is likely to be the last edit. Jump to error seems to move the cursor to the start of the (sometimes next) line, breaking flow if you make a typo.

Expected behavior If the error is near the cursor, don't move it.

Desktop (please complete the following information):

Additional context Disabling jump to error globally is not desirable. Maybe column number could be extracted from shader error log?

3Dickulus commented 3 years ago

Column number is a good idea.

I've found that getting the line right line number is problematic, +/- 1 is caused by QOpenGLShaderProgram injecting lines into the sources. I tried my best to compensate but it doesn't seem consistent across GLSL versions ??? the result is that sometimes the line number is out by one. Could also be the result of me not quite understanding exactly why and when Qt injects extra lines.

3Dickulus commented 3 years ago

the GLSL compiler reports column? eg:Examples/Historical 3D Fractals/Mandelbulb.frag (90) : error C7011: implicit cast from "float" to "int" I see only line number only as reported from shaderProgram->log()

claudeha commented 3 years ago

Here with:

  return c; error;

on line 68 (as reported by the editor line numbers) I get

/home/claude/FractalHelix-2.frag :69(12): error: `error' undeclared

The 12 is the column number (e of error), and the line number is off by one. Until the offset is fixed I suggest making "nearby" be fuzzy (abs(x-y)<=d rather than x==y).

I'm using Debian Bullseye (current testing):

Vendor: X.Org
Renderer: Radeon RX 580 Series (POLARIS10, DRM 3.39.0, 5.9.0-3-amd64, LLVM 11.0.0)
GL Driver: 4.6 (Compatibility Profile) Mesa 20.2.3
3Dickulus commented 3 years ago

yup, fuzzy can be added if needed, I would rather figure out what is causing the off by one, can you post me the frag pak? auto saved to _Files/ with include support frags, for the most part my system does not show off by one, if those same frags show the same off by one on nVidia as on Radeon then manufacturer differences could be ruled out and I'll dig into FragM sources some more.

edit: ...and I'll poke at the glsl compiler a bit to see where the column number is hiding...

3Dickulus commented 3 years ago

log lines look like this on my system 0(89) : error C7011: implicit cast from "int" to "float" leading "0" is nVidia's glsl compiler file index number generated by #include directives replaced with filename by FragM, no column numbers :(

claudeha commented 3 years ago

Opened a new issue specifically about the off-by-one problem: #157 I still think this fuzzy match is useful even when the off-by-one is fixed, thanks for the patch, will try it soon.