64bit and fixes #271
64bit and fixes #271
Conversation
…_t, this avoids crashes. Still renders glyphs incorrect
memmove allows overlapping ranges. But there might be an actual bug.
…issuu-64bit-andfixes
* master: (26 commits) Resolve merge conflict with main branch. [buildCFF2VF] REsolve merge conflict with main branch [buildCFF2VF'. Fix bug with builder.buildVarData() which blocks building a CFF2 VarStore [buildCFF2VF] Added work-around for bug in fonttools [buildCFF2VF] Fix support for CFF2 fonts with multiple FontDicts [buildCFF2VF] Remove debug line [buildCFF2VF] Added work-around for bug in fonttools [buildCFF2VF] Fix bug in building from name-keyed CFF masters [blendCFF2VF] Assign a value of False rather than 0 to boolean fields blendCFF2VF] Fix PEP8 conflicts: line length, white-space. [buildCFF2VF] Fix Codacy error for unused operator. [buildCFF2VF] Fix flake8 error reports [buildCFF2VF] Fix support for CFF2 fonts with multiple FontDicts [travis] Replace bare Slack token [makeotfexe] Remove code for preventing idDelta overflow in cmap format 4. makeotfexe] Fix issue #242, Unnecessary truncation of the Format 4. [checkoutlinesufo] Change --version option to print only the tool's version number [checkoutlinesufo] Fix __doc__ string. Move tool description to the top. Update Codacy badge links Correct change made in c55de1b ...
A mistake in the format specifier for reading Unicode values from the GOADB file made the Unicode assignment not work at all. Comparing output between the 32 bit and 64 bit makeotfexe led to finding a very old bug with un-initialized variables when filling mark to base records. Anchor records were not fully zeroed when created, resulting in a failure to match two records that should match. Old fonts are correct, they just sometimes have more anchor records than needed.
|
@readroberts why re-indenting whole files when fixing the affected lines would have been simpler and would not introduce a bunch of unnecessary changes? This PR is messed up too much for my taste. |
|
@msousa I choose to re-indent the entire file because there are many inconsistencies in indentation and spacing throughout the file. If I am fixing one spot, why not make them all consistent? I note that the only changes in the last commit are whitespace. I don't care that much about the issue - if you really don't want the entire file re-indented, I can revert the change, let me know, and I will fix only the spots that recently had tabs converted to spaces by Xcode. |
|
@readroberts changing just the lines I marked would have been a lot better IMO. Please undo the changes by rewriting the commits instead of reverting. |
…r a variable. Set build architecture back to 32 bits.
readroberts
force-pushed
the
issuu-64bit-andfixes
branch
from
85ec115
to
7a6376f
Compare
…nvironment. Add references to <stdint.h>, and make function declaration and definition use the same types.
… won't fail on this branch
Fixed operator precedence issue by adding parents to DF_LEVEL in feat.h, as pointed out by Chris Chapman Fixed several printf format warnings about signed/unsigned, long/int differences between arguments and specifiers.
…intf format specifier match with arguments
|
I have now installed cppCheck - required fixing some stuff with Xcode. I recently installed High Sierra, but I had some old stuff that interfered with Xcode 9. cppCheck shows many warnings about makeotf - over 500, with '--enable=warning --enable=style' - but this commit fixes all the issues with the change from 32 to 64 bit. I propose that this PR should be merged with main branch now, and that the many cppCheck non-consequential style and printf format warnings should be pursued as a separate project. |
This is the revival of PR #161