logfile.txt
Running the test-suit on the un-modified committed etcpack.exe binary shows failed tests.

An un-modified re-compile produces completely different faults.
And many many more of them.
Was the binary in master compiled from master?
recompiled_logfile.txt

Windows-10 x86_64.
Visual Studio 2008. Version 9.0.30729.1 SP

I want to know how to compile it on macOS? Thanks.

When I use gcc etcpack.cpp to compile it, I got a error:

etcpack.cpp:15573:6: note: use '==' to turn this assignment into an equality comparison
        if(f=fopen(dstfile,"wb"))
            ^
            ==
etcpack.cpp:15885:16: error: variable has incomplete type 'struct _timeb'
        struct _timeb tstruct;
                      ^
etcpack.cpp:15885:9: note: forward declaration of '_timeb'
        struct _timeb tstruct;
               ^
59 warnings and 1 error generated.

Hi;
Are you planning on licensing this with more permissive license such as MIT?

If I understand correctly the current license seems to prohibit redistribution if any modifications (such as fixes, enhancements or 'glue code') are made which is a likely showstopper for projects using this code.

// Tapani

is there a version where it's already built? I want to decompress etc.dds files but I don't have coding experience at all :(

The following section is unclear from a legal standpoint:

"Subject to the terms and conditions of the SLA, the licensee of the
Software (the "Licensee") hereby, receives a non-exclusive,
non-transferable, limited, free-of-charge, perpetual and worldwide
license, to copy, use, distribute and modify the Software, but only
for the purpose of developing, manufacturing, selling, using and
distributing products including the Software in binary form, which
products are used for compression and/or decompression according to
the Khronos standard specifications OpenGL, OpenGL ES and
WebGL. Notwithstanding anything of the above, Licensee may distribute
[etcdec.cxx] in source code form provided (i) it is in unmodified
form; and (ii) it is included in software owned by Licensee."

From my understanding, if we want to distribute the source code, it needs to be distributed as-is with the same license. If we want to change the code, we need to distribute it in binary form.
What about distributing it in binary form without doing compression and decompression? What about products that do compression/decompression that does not follow the Khronos standard specification?
What about a product that is actually another library A, where another user uses this library to build another library B or an application C and does not use the compression/decompression capability of the original library A in library B or application C?

In which circumstances we cannot redistribute the Software in a binary form?

Thank you!

I use Visual Studio 2015 compile the code in VC9. When I compare the logFile.txt. I found some results are not same.
logfile.txt

First, is this project active? I see a recent change four months ago.

However, upon trying your etcpack.exe I encountered and error because my Windows 10 system does not have "imconv". I investigated further and learned that imconv is an image conversion tool which was last updated in 1995. I had to dig quite a bit to find the source for this tool as many links from 1995 do not work.

ftp://ftp.sdsc.edu/pub/sdsc/graphics/imtools/source/

Given that this tool has not been updated in over 20 years, and ImageMagick was updated yesterday, perhaps the etcpack source could be updated to reflect a modern image conversion tool? Presumably, this would save many people, myself included, the hassle of having to make the change themself and rebuild from source.

Per the KTX specification, the glBaseInternalFormat field should never be a typed field with bit widths baked into the identifier.

The etcpack executable bundled with the Mali Texture Compression package reports:
ETCPACK v4.0.1 for ETC and ETC2

And the code in this GitHub repo reports:
ETCPACK v2.74 For ETC and ETC2

I cannot find the source code of the newer version anywhere online. The new version is claimed to be much faster per Mali Texture Compression webpage:

Improved ETC1 compression speed of up to 600x compared to ETCPack at the same or better quality.

(https://developer.arm.com/tools-and-software/graphics-and-gaming/mali-texture-compression-tool)

Has etcpack become closed source?

etcpack.cxx (possibly other files) extensively uses static enum. This is not valid C/C++ (and doesn't make any sense). static determines how storage should be allocated but an enum-declaration does not allocate storage.

This means it compiles in MSVC but not in MinGW (I assume not gcc as well, or MSVC when using /permissive-).

The solution is to simple replace static enum with enum.