Bug introduced in b3ca003
Fixes #350

Fixes #338

Found these by running 'python3 -m pytest'

https://github.com/fonttools/fonttools/blob/master/Lib/fontTools/ttLib/ttFont.py#L311
Bug introduced in f72057f

Passing values around was getting too unwieldy for my taste

AppVeyor is getting hung on them, don't know why

This is needed for diff'ing the PDFs that are generated by the *plot tools. By default each page contains a header (a.k.a. Title) that includes a time stamp, among other info. Each PDF page is then compressed using ASCII85. By removing the page header it's possible to do a straight diff of two PDFs, even if their pages are compressed.

In case raw (i.e. uncompressed) pages are needed, use the setPageCompression() method of pdfgen.py

otherwise stems are written out with the default values when we want blended stems

* digiplot integration tests

* Remove 'CID charsets' folder

* Remove code related to 'CID charsets'

[ttxn] Fix shell calls to 'tx'

Fixes #337

instead of `master`

[skip ci]

This shall be attempted again at a later date. For more details see d64f22f#commitcomment-29108842

With the 64-bit update, the function global.c::getGlyphName() can return a glyph name of up to 128+7 characters long. The returned string was, in many places, copied into a string buffer that was much shorter.

[skip ci] Fixes #328

[skip ci]

* [tx][varread] Fix issue #272: Fix inconsistency of use of regionCount variable in tx.varread.varread.c

This update makes no change in function: it simply renames variables and fields to avoid confusion by developers.

* [tx][tx_test.py] Add test for reading and snapshotting simple CFF2 VF font to test fix for issue #272

(we will let the pyup bot take care of updating requirements.txt)

* In differ.py, when it walks a dir tree to make a list of files in a ufo file, the list needs to be sorted to avoid the difference between the reference directory tree and the new temporary director tree when the test is run under Windows. The lists were compared, and the diff was failing before the files were even examined.

* In tx_test.py, when pdf files are compared, an additional line in the pdf file needs to be ignored. This didn't cause a problem before, but as of June 1st, the date string changed enough to make this extra line change.

* Replaced tmpfileNameTempFile() with tmpfile.mkstemp(). The former not only creates a file, but also opens it. This appears to conflict with writing to the same file path from other processes. In any case, this change fixes a lot of failures that appears on some, but not all, Windows systems.

* Use absolute path for tx.exe

* Fixes for Test under Windows with ttxn. Referencing the absolute path to tx.exe fixes a failure when running the test under Appveyor.
Saving the output from a shell call to tx.exe to a file name rather than a tmpfile.TempFile file object fixes a failure under Windows OSX 10 when running pytest manually.

* Shift to consistent use of tempfile.mkstemp() instead of NameTempFile().

* Add test for when command fails. Note that I did not handle the 'subprocess.CalledProcessError' in runner.py::main(). I notice that all the exceptions that can be raised by the argparse module are not handled, and there is no exception handling for any other exception. I think it is simpler to let the 'subprocess.CalledProcessError' pass through and be caught and reported by pytest.

* Add some debug code to show path to tool when there is a CalledProcessError.

* Remove type1 and detype1 tests that run the programs on a file of their intended output file type as input. The programs are not designed to fail in any reliable way in this condition. The current tests cause detype1 to fail with the same error on Mac and Windows, but do not cause  type1 to fail.

This reverts commit 2d6ee43.

…ters

... to build only the release projects if no target is given.

They will also run the right build for the targets 'release', 'debug', and 'clean', and do nothing for any other target.

Ilya Finkelshteyn said that the code removed by this commit is no longer needed now that appveyor/ci#2066 is fixed.
(This is round 2, because I'm trying again off a fresh branch from `develop`.)

Fixes #417

Possible fix for https://ci.appveyor.com/project/adobe-type-tools/afdko/build/1.0.432
Traceback (most recent call last):
  File "c:\python27-x64\lib\site-packages\afdko\Tools\SharedData\FDKScripts\MakeOTF.py", line 3062, in main
    runMakeOTF(makeOTFParams)
  File "c:\python27-x64\lib\site-packages\afdko\Tools\SharedData\FDKScripts\MakeOTF.py", line 2736, in runMakeOTF
    os.remove(outputPath)
WindowsError: [Error 32] The process cannot access the file because it is being used by another process

we need to skip the step where python3's distutils attempts to fix up the
script's shebang, otherwise it fails with a SyntaxError, since it expects
them to be text files formatted as python code, whereas in our case they
are simply C native executables.

changed OpType from enum to a typedef (of an enum)
(also, now explicitly using OpType instead of int in OpRec struct)

These two tools are mostly a copy-paste of tx

This project has *.txt files that used to be the source for the *.h in-line help files. That was years ago, and we don't even have the tool that built the *.h from the *.txt files anymore. The *.h files are now the master source for the in-line help text.

…iles

related to commit 6e88dd8

* Pin fonttools to latest version 3.27.0

* change pyup's `==` to `~=` for fontTools