WRL upload changed for the worse ?

Discussion in 'Bug Reporting' started by mkroeker, Oct 23, 2012.

  1. mkroeker
    mkroeker Well-Known Member
    Hi, I hate to ask this, but has something changed in the VRML2 upload machinery (beyond the improved
    support for texture file paths mentioned a few days ago) ?
    I have a couple of rather simple files that I uploaded on Oct 2 which were subsequently printed without apparent
    problems - except that I had made a silly mistake in one of them. Re-upload of the fixed file surprisingly ended
    with the dreaded "uh-oh" message, and during testing I now find that neither of the original, already printed
    wrl datasets will upload anymore. One example is http://shpws.me/leS7 , which I have now made both
    public&downloadable to facilitate testing.
    I do not see anything special with these files - they contain a bunch of VRML "Sphere" and "IndexedFaceSet"
    primitives with embedded per-shape RGB colors. Uploading a single red sphere still works, but I must admit
    that I have not tried a colored cube yet. (Assuming that this is somehow related to the recent rework of the
    color-by-texture-file code at all)

  2. MitchellJetten
    MitchellJetten Shapeways Employee CS Team
    Hi Martin,

    My apologies for not replying earlier, on the background we have been working on this issue.
    Our development team has confirmed this issue and started working on a fix the moment they read your post.

    If everything went alright, this should now have been fixed!
    " Triangle count for vrml model with sphere primitive explodes after processing"

    Could you let me know if you are still having issues?
  3. mkroeker
    mkroeker Well-Known Member
    Works perfectly fine again, thank you very much.
    The obvious drawback is that I now have another bunch of models to order, including a nicer
    version of what is in my current order (bloody expensive hobby, this :blush: ).
    l really should have waited for you to work your magic instead of trying clumsy workarounds
    at the border of non-2-manifold land.
    Would it be possible to pass back more specific error codes to the thing that sends the rejection
    email ? "Polygon count exceeded" would be so much more helpful than "model needs more love"
    (OTOH, if it were easily doable you would probably be doing it already...)