›

Diff tool does not use XOR

TheoTheo
edited March 2016 in KLayout Support

I have two large gds database on which I run diff with the option "run XOR on differences"
A lot of differences are reported, but I do not see what is different. When I pick a cell and run XOR on that cell there are indeed no differences. This makes me believe that the diff tool does not actually run an XOR.
If run XOR on the entire database (only some visible layers) it reports also no differences. The gds are created by different tools so some differences could be present due to the writer. Any idea?

Comments

  • MatthiasMatthias
    edited November -1

    Hi Theo,

    that's hard to tell without a sample.

    In general, the diff tool tries to identify "equivalent" cells. It uses several ways to identify cells - by name, by their position, by their size - to be able to catch renaming scenarios. This heuristics may fail and the tool will report a lot of differences.

    In addition, running XOR's on single cells may give false positives (in the sense of differences) even though finally (on top level) no real differences exist. For example, if a cell is placed rotated in one file and in the other file, the cell is not placed rotated, but the content is rotated, the cells will differ, but on top level the geometry appears the same.

    Differences may also be reported for texts which do not show up in the XOR.

    What kind of differences do you see?

    Matthias

  • TheoTheo
    edited November -1

    All cells have the same size and rotation. I know one of the tools breaks up polygons or smashes paths to shapes and there is some cell renaming. That I expect to give a difference but nothing with an XOR. So it should work as I had expected.

    If I choose a very small cell as top level (with a reported difference) I get the same result. For instance a difference is marked for a layer that is not being used in both cells. There is however a difference, but that's on another layer. So I think the problem is that the tool point to the wrong layers. I tried to make a testcase by only exporting these two cells, but it gives a different result. The XOR reports 4 different layers (correct 2 layers have been remapped) while diff only reports one layer difference.
    I can send you this small testcase if you want?

    Theo

  • MatthiasMatthias
    edited November -1

    Thanks for sending the test case. I was able to locate and fix the issue. It was related to different layers in the two files - this case was not handled properly and difference markers appears where they should not.

    The fix will be released in the next minor release.

    Thanks and best regards,

    Matthias

  • TheoTheo
    edited November -1

    Thanks Matthias

    The tool only becomes better!

  • MatthiasMatthias
    edited November -1

    Hi Theo,

    0.24.6 is released now which should fix this issue.

    Kind regards,

    Matthias

Sign In or Register to comment.
Powered by Vanilla