Hyogeol Lee

I attached compiler error log by VC6. But it seems hard to fix build errors. From error log file, you can find - VC6 does not support unsigned long long....

Hi, I found different size of checksum in AVIOContext. So it has different offset after checksum. https://github.com/Ruslan-B/FFmpeg.AutoGen/blob/5194d118f8e8f4e0f3ae83ffd311aacb2d0f1c66/FFmpeg/include/libavformat/avio.h#L246 https://github.com/Ruslan-B/FFmpeg.AutoGen/blob/d7f6e2309a46f0814067152ac1ce58aa84abe251/FFmpeg.AutoGen/FFmpeg.structs.g.cs#L2060

In FFmpeg.Autogen, size of checksum is 8 bytes (ulong = uint64). But in native version, size of checksum is 4 bytes in common case (unsigned long = uint32). So it...

In 4.x, there was no critical fields around seekable in my guess. https://github.com/Ruslan-B/FFmpeg.AutoGen/blob/e0d26f8a24e36b6cec25c703336eddb38a6d017e/FFmpeg/include/libavformat/avio.h#L260 And you need to consider size of unsigned long by platforms. In .NET, ulong is always 8...

Would you binarize your image manually?  ``` .\tesseract.exe -l jpn_vert --psm 5 binary.png - 減る記憶、 それでも増える、 パスワード ```

It is hard coded, you need to write new number parser. https://github.com/stephan-tolksdorf/fparsec/blob/156cbd751fac67ca7cc04cf661710f1f90a216a5/FParsec/CharParsers.fs#L996 https://github.com/stephan-tolksdorf/fparsec/blob/156cbd751fac67ca7cc04cf661710f1f90a216a5/FParsec/CharParsers.fs#L1153 https://github.com/stephan-tolksdorf/fparsec/blob/156cbd751fac67ca7cc04cf661710f1f90a216a5/FParsecCS/HexFloat.cs#L233

Maybe related to https://github.com/dotnet/coreclr/pull/11019 And it seems not ported to .NET framework runtime.

> Hey [@hglee](https://github.com/hglee) , thanks for bringing up the issue. > > Would it be possible to share which profiler you're using for the benchmarking? Hi, it's Jetbrains dotTrace (https://www.jetbrains.com/profiler/)....

You can avoid it by filling a email to the user settings in the form and saving it. It's not a bug: https://github.com/freerouting/freerouting/commit/bb9cf14154f4dbfd0fe4698297b62cb357c7b761

It seems like the input needs to be converted to a gray image before. Some backends do the conversion internally, but the ffmpeg backend seems to require the conversion externally....