Readable shader error messages #2622
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Nice! @bors-servo r+ |
|
📌 Commit fc8a16c has been approved by |
bors-servo
pushed a commit
that referenced
this pull request
Apr 5, 2018
Readable shader error messages Working with shaders, I find `#line` navigation experience to be rather poor: 1. first, I take an `apitrace` capture 2. load the capture, go to the last `glShaderSource` 3. copy out the input into a new file 4. for each `#line` directive, try to match the reported error number ... This PR removes line overrides. Instead, it prints the relevant piece to the console, e.g.: ``` Loading shaders... Failed to compile shader: cs_clip_image 0:1152(17): error: syntax error, unexpected NEW_IDENTIFIER, expecting ',' or ';' | ClipScrollNode scroll_node = fetch_clip_scroll_node(cmi.scroll_node_id); > DImageMaskData mask = fetch_mask_data(cmi.clip_data_address); | RectWithSize local_rect = mask.local_rect; ``` This information instantly gives me an idea of what's wrong (note: unlike the original driver message), and makes all the listed steps unnecessary for 99% of cases. And when the steps are necessary, we don't have to check for each `#line` occurrence since the driver-reported line number is no longer ambiguous. <!-- Reviewable:start --> --- This change is [<img src="https://reviewable.io/review_button.svg" height="34" align="absmiddle" alt="Reviewable"/>](https://reviewable.io/reviews/servo/webrender/2622) <!-- Reviewable:end -->
|
☀️ Test successful - status-appveyor, status-taskcluster |
bors-servo
pushed a commit
that referenced
this pull request
Apr 6, 2018
Separate primitive shaders from clip shaders Includes #2622 Note: this is not done because I'm bored. The problem Szeged team has bumped into is that `aDataX` attributes end up being used in the shaders they don't relate to (e.g. clip shaders). This isn't affecting GLSL but it does affect GLSL -> SPIRV -> XXX shader pipelines. The solution I came up with involves moving parts of `prim_shared` into separate shared modules, and turning `clip_shared` into a catch-all header of clip shades. Downsides: more shader modules (but not programs) to navigate between. Any suggestions on alternative solutions are welcome ;) <!-- Reviewable:start --> --- This change is [<img src="https://reviewable.io/review_button.svg" height="34" align="absmiddle" alt="Reviewable"/>](https://reviewable.io/reviews/servo/webrender/2623) <!-- Reviewable:end -->
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Working with shaders, I find
#linenavigation experience to be rather poor:apitracecaptureglShaderSource#linedirective, try to match the reported error number ...This PR removes line overrides. Instead, it prints the relevant piece to the console, e.g.:
This information instantly gives me an idea of what's wrong (note: unlike the original driver message), and makes all the listed steps unnecessary for 99% of cases. And when the steps are necessary, we don't have to check for each
#lineoccurrence since the driver-reported line number is no longer ambiguous.This change is