hi before asking my question i want to thank shree whose comments are very helpful both here and in github repo of tesseract.
i want to fine tune fas.traineddata to support some new fonts. the first problem arises when i use the following command:
tesstrain.sh --fonts_dir /c/folder/fonts/ --lang fas --noextract_font_properties --linedata_only --exposures "0" --langdata_dir ../langdata --tessdata_dir ../tessdata --fontlist "b nazanin" --output_dir ../../tessdata/fas/

i put fas.traineddata, which i downloaded tessdata_best repo, in ../tessdata folder, but it gives error and says that it can not find eng.traineddata. this problem is resolved when i put eng.traineddata in ../tessdata but why should it want eng when i specify that lang is fas?
anyway for now i pasted eng,traineddata and moved on. the second problem is with tiff/box pair generated with the above command. first of all in some words in tiff files the characters are not joined.for example there is:
but it should be
another problem is that the box file generated is from left to right but it should be RTL. this problem is addressed here but i did not understand if there is a solution for it or not.
lastly i am confused with the fine tuning process. is tesstrain.sh only for generating tiff/box pairs? what are the next steps. is using lstmtraining.exe the next and final step?
btw i'm using:
tesseract 4.0.0-beta.3
leptonica-1.76.0 (Jul 10 2018, 21:36:38) [MSC v.1900 LIB Debug x64]
libgif 5.1.4 : libjpeg 9b : libpng 1.6.34 : libtiff 4.0.9 : zlib 1.2.11 : libwebp 0.6.1 : libopenjp2 2.3.0
Found AVX
Found SSE
which i built with vs2015 also i'm using win 8.1