Papaya segfaults on Gentoo

[maffblaster]

Hi @ApoorvaJ,

I successfully built Papaya on the current HEAD (744f658), however when trying to open it I'm getting the following message:

Attribute UV not found in unlit.glsl
Uniform Texture not found in shader at unlit.glsl
Segmentation fault (core dumped)

When I check out the tagged version (v0.1) it builds and runs, however it seems functionality is quite limited. What suggestions do you have to resolve this issue on current HEAD?

One other question: does Papaya require an Nvidia graphics card in order to operate? I noticed it was dynamically linked against quite a few Nvidia .so's :)

[ApoorvaJ]

Thanks for the bug report. There are a few issues at play here:

The unlit.glsl errors
This shader is only used for displaying the crop outline, and as such this error does not affect functioning of the shader. I do plan to fix it, but am working on other stuff at the moment.

The NVIDIA linkage
Papaya does not depend on any NVIDIA-specific functionality. I'm not sure about how the NVIDIA-specific .so file dependencies got there, but they might just be the result of other libs e.g. GTK, on your machine. I have tested this on Intel and AMD machines as well, and it works.

The SegFault
I'd expect the HEAD to be broken in some way for quite some time now, since I am working on the core infrastructure behind layering and effects, and this system interacts with every other component in Papaya. Until I get this in place, all the other stuff will be broken. This is one of those cases where Papaya will get worse until it gets better. :)

Your best bet is to use v0.1, or roll back to an older commit in the git history and build. To be honest, though, Papaya is not even remotely ready for real-world usage at this point. I do have a clear picture of the road ahead now, and things should improve.

Keeping this bug open until things stabilize again at HEAD.

[maffblaster]

@ApoorvaJ, thanks for all your hard work! I'll wait until you fix these things and try again. Do you have a specific milestone for the next tagged version?

[ApoorvaJ]

I'm not sure about a tagged release, but things should stabilize when I have layering working with the bitmap layer, any one effects layer (most likely hue/saturation), and when the node window is working. I'm currently working on basically all three at the same time, in order to figure out the API design.

[k2t0f12d]

e210fd6 is the newest that builds for me

[ApoorvaJ]

As I said, stuff shall be broken for a while. I'm currently bringing up the node API on Linux, so builds will fail on Windows for a while. Builds should succeed on Linux, but may segfault.

I'm pushing all the instability to the master branch because multiple branches are completely unnecessary at the moment, since we don't have any external code dependencies or release schedule.

[k2t0f12d]

Its all good, just put that in here for someone, like myself, who might like to find a version that was still working. They can go directly to that commit and build.

[ebachard]

Hello,

Thanks for your interesting work !

I don't have much time, but if this can help a bit, I confirm :

• Papaya does build without a glitch on Linux (Debian like, g++-5.2.1)
• ... and gives a segfault, with the associated message :
  "Attribute UV not found in unlit.glsl
  Uniform Texture not found in shader at unlit.glsl'

Reading the code, looks like all the images path are hard coded. Is it normal ?

In fact, I think there are several mied issues:

1. path issues

egrep -Hr define ./* | grep apoorvaj
./linux_ui.cpp:#define PAPAYA_DEFAULT_IMAGE "/home/apoorvaj/Pictures/o1.png"

( + other similar wrong path in several files)

This is very probably (one of the) segfault root ...

1. maybe a wrong shader code in unlit.glsl

Currently :
gl_FragColor = Frag_Color;

In other .glsl files, we got:
gl_FragColor = Frag_Color * texture2D( Texture, Frag_UV.st);

If I find mode time, I'll investigate. Suggestions : use relative path for images, and e.g. copy all the images beside the shaders in the build/linux directory ? (just a wild suggstion, maybe I'm completely wrong)

Thanks again for sharing your code :-)

ericb

[ApoorvaJ]

Ericb, I know both these points of failures. As I said before, I'm working on other stuff and the master branch is very unstable at the moment. No need to look into this further.