Re: [NEW] VeraCrypt 1.25.4 with GUI, with upstream support

Omar Polo <op@openbsd.org> writes:

> Tietoturvamies <main@tietoturvamies.fi> writes:
>
>> Hi,
>>
>> the patches by jcs adding OpenBSD support into VeraCrypt were accepted
>> upstream [1] and the new release [2] does include them, so
>
> that's good to hear!
>
>> I've updated the port and it's attached.
>>
>> Changes to previous port [3]:
>> - no additional files or patches are needed now, everything is in the
>> upstream as recommended by Porting Guide
>> - updated maintainer from my previous e-mail to current one
>> - versions harakiri not needed for this version as there is no
>> update/Update confusion in URLs in upstream
>>
>> Caveats:
>> - only FAT is supported, unlikely to change due to the port being "more
>> or less just for compatibility reasons"
>> - sudo is used instead of doas
>>
>> If somebody is interested, can you give it a try, check if it works for
>> you and give some feedback?
>>
>> Any change the port could then be imported into ports tree? I'd like to
>> be a maintainer, unless somebody else wants to be one.
>>
>> Thanks,
>> TTM
>>
>> [1] https://github.com/veracrypt/VeraCrypt/pull/779
>> [2] https://github.com/veracrypt/VeraCrypt/releases/tag/VeraCrypt_1.25.4
>> [3] https://marc.info/?l=openbsd-ports&m=160259246413274&w=2
>
> some comments:
>
> - REVISION=0 is not needed, bumps to that are for when a port is
> changed after being imported.
>
> - I'd keep DISTNAME closer to PKGNAME at the top as per
> Makefile.template
>
> - nitpicking but WANTLIB are usually indented differently
>
> - your tarball had wrong permissions that wouldn't allow the build
> using PORTS_PRIVSEP.
>
> - upstream tarball also needs FIX_EXTRACT_PERMISSIONS=Yes to build.
>
> - the build process hides all the command executed, this make
> impossible to see if it picks up CFLAGS etc correctly. I'm quite
> puzzled by this, as Build/Include/Makefile.inc which contains the
> rules doesn't use '@' to hide their output, and I don't see any
> -s/--silent flag passed to gmake.

Thanks Stuart for tracking this down. It needs VERBOSE=1, and then some
massaging to pick up "our" CFLAGS/CXXFLAGS, see patch-Makefile. Adding
a custom PORT_CFLAGS/CXXFLAGS variable seemed the quickest way to inject
them, upstream makefile is quite complex and I lost track of the status
of the variables between all those export .. :=, +=, ?= etc...

> - The port (probably) uses C++11, so add a comment about that before
> the COMPILER line and add ports-gcc to the list.
>
> - I'd add NO_TESTS=Yes since there doesn't seem to be a regress suite.
>
> - why not install the desktop file too? I went with a patch for
> veracrypt.desktop instead of some sed magic, but either works. There
> are some xmp icons, they work with jgmenu but I don't know if they're
> widely supported.
>
> otherwise seems fine, the GUI pops up and I can click around. I haven't
> really done any serious runtime testing thought.
>
> attaching a tarball with the above points addressed.
>
> Cheers,
>
> Omar Polo