Revert "[fix] MuPDF Android font dir patch update (#971)"

[Frenzie]

This reverts commit b7dd4c1.

Closes koreader/koreader#5617.

---

This change is 

[Frenzie]

@NiLuJe I'm a bit confused here. There's something happening only on Cervantes… unrelated to this PR of course so I'll go ahead and merge it now, but pretty weird.

make[3]: Leaving directory '/home/ko/base/thirdparty/fbink/build/arm-cervantes-linux-gnueabi/fbink-prefix/src/fbink'

make stripbin

make[3]: Entering directory '/home/ko/base/thirdparty/fbink/build/arm-cervantes-linux-gnueabi/fbink-prefix/src/fbink'

fatal: No names found, cannot describe anything.

arm-cervantes-linux-gnueabi-strip --strip-unneeded Release/fbink

make[3]: Leaving directory '/home/ko/base/thirdparty/fbink/build/arm-cervantes-linux-gnueabi/fbink-prefix/src/fbink'

make[2]: Leaving directory '/home/ko/base/thirdparty/fbink/build/arm-cervantes-linux-gnueabi/fbink-prefix/src/fbink'

/home/ko/base/thirdparty/fbink/build/arm-cervantes-linux-gnueabi/fbink-prefix/src/fbink/tools/do_debian_package.sh Release armel

Building FBInk--armel.deb

dpkg-deb: error: parsing file 'Release/pkg/DEBIAN/control' near line 2 package 'fbink':

 error in 'Version' field string '': version string is empty

Makefile:535: recipe for target 'cervantes' failed

make[1]: *** [cervantes] Error 2

make[1]: Leaving directory '/home/ko/base/thirdparty/fbink/build/arm-cervantes-linux-gnueabi/fbink-prefix/src/fbink'

ninja: build stopped: subcommand failed.

Makefile.third:484: recipe for target 'build/arm-cervantes-linux-gnueabi/fbink' failed

make: *** [build/arm-cervantes-linux-gnueabi/fbink] Error 1

[NiLuJe]

The Cervantes target tries to build a Debian package, that bit is normal. As to why it fails, no idea... Ping @pazos ;).

…

On Thu, Nov 21, 2019, 14:26 Frans de Jonge ***@***.***> wrote: Merged #1013 <#1013> into master. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#1013?email_source=notifications&email_token=AAA3KZQVQZQR5RWQZGCGBYTQU2EBTA5CNFSM4JQAU3WKYY3PNVWWK3TUL52HS4DFWZEXG43VMVCXMZLOORHG65DJMZUWGYLUNFXW5KTDN5WW2ZLOORPWSZGOVAICRCQ#event-2819631242>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAA3KZSHXBAEUY2ZO6UA2MLQU2EBTANCNFSM4JQAU3WA> .

[NiLuJe]

Oh, unless... Because it's no longer pinned on a tag, but on a random commit.

I'll look into it (and tag a release soon anyway ;)).

[NiLuJe]

Ah, no, it's simply that git describe requires an unshallow clone to behave properly, I think?

[NiLuJe]

It's also highly unclear to me how its error message (fatal: No names found, cannot describe anything.) managed to make it to your log despite a 2>/dev/null... ^^

[Frenzie]

git describe has to be able to reach a tag, but the script I wrote only goes deeper than 50 if it fails to checkout the tag or commit. If you've got more than 50 commits since the last tag maybe it's about time? :-P

4b8bc32

[Frenzie]

Anyway, if it were purely about git describe then surely it should happen just as much on Kobo?

[NiLuJe]

Yeah, switching to a tag forces an unshallow clone since master is currently 55 commits away from the last tag ;).

That particular visible git describe call is Cervantes-specific, but there's another one buried in the Makefile, which is currently probably doing something very stupid, like using the error message as a version string :D.

[NiLuJe]

There, fixed it ;).

Plus, I'll tag & bump a release soon anyway ;).