podofo @0.9.4 +universal: non-const lvalue reference to type 'pdf_long' (aka 'int') cannot bind to a value of unrelated type 'long'

[ryandesign (Ryan Carsten Schmidt)]

test/unit/EncryptTest.cpp:211:76: error: non-const lvalue reference to type 'pdf_long' (aka 'int') cannot bind to a value of unrelated type 'long'
        pEncrypt->Decrypt( pEncryptedBuffer, nOutputLen, pDecryptedBuffer, m_lLen );
                                                                           ^~~~~~
src/base/PdfEncrypt.h:371:59: note: passing argument to parameter 'outLen' here
                         unsigned char* outStr, pdf_long &outLen) const = 0;
                                                          ^
1 error generated.

[ryandesign (Ryan Carsten Schmidt)]

Builds fine when not using the universal variant.

[dbevans (David B. Evans)]

Bummer, I was really feeling good about finally fixing the "can't build with old installed copy" problem. But I'll take a look. My experience is that cmake and +universal don't always work well together. Too much going on silently (and often undocumented) behind the scenes with cmake. Thanks for catching this.

[ryandesign (Ryan Carsten Schmidt)]

Replying to devans@…:

Bummer, I was really feeling good about finally fixing the "can't build with old installed copy" problem.

That's not it: this happens even if podofo is not already installed.

My experience is that cmake and +universal don't always work well together.

This particular problem isn't related to universal; it's related to building 32-bit. podofo @0.9.4 doesn't build single-arch i386 either; log attached.

[dbevans (David B. Evans)]

Replying to ryandesign@…:

Replying to devans@…: This particular problem isn't related to universal; it's related to building 32-bit. podofo @0.9.4 doesn't build single-arch i386 either; log attached.

I was just coming around to that way of thinking myself. Thanks for the logfile. There may be hope yet.

[dbevans (David B. Evans)]

Fix committed in r153780.