From: izzy Meyer <izder456@disroot.org>
Subject: Re: NEW: devel/msitools at 0.106
To: "Anthony J. Bentley" <anthony@anjbe.name>
Cc: ports@openbsd.org
Date: Mon, 22 Jun 2026 14:28:08 -0500

On Mon, 22 Jun 2026 09:43:09 -0600
"Anthony J. Bentley" <anthony@anjbe.name> wrote:

> izzy Meyer writes:
> > > Attached is a fairly simple port of msitools. I had to use this to
> > > play a windows game made in rpgmaker2000 on OpenBSD. Worked fine
> > > for extracting that msi file. Could be useful for other people on
> > > OpenBSD, probably.
> 
> ===>  Configuring for msitools-0.106
> The Meson build system
> Version: 1.11.1
> Source dir: /usr/ports/pobj/msitools-0.106/msitools-v0.106
> Build dir: /usr/ports/pobj/msitools-0.106/build-amd64
> Build type: native build
> Project name: msitools
> Project version: UNKNOWN
> C compiler for the host machine: cc (clang 22.1.6 "OpenBSD clang
> version 22.1.6") C linker for the host machine: cc ld.lld 22.1.6
> Vala compiler for the host machine: valac (valac 0.56.19)
> Host machine cpu family: x86_64
> Host machine cpu: x86_64
> Program perl found: YES (/usr/bin/perl)
> Program bison found: NO
> 
> Added bison, and then got:
> 
> Program bats found: NO
> 
> So:
> 
> - add BUILD_DEPENDS on bison/bats
> - run 'make update-plist' to trim some unnecessary directories from
> PLIST
> - might as well enable debug packages (IMO)

Thanks for the nits, attached is a new port with these fixes. With
these changes, can I get OKs?

> - any idea why 'make test' fails?

from the log file meson informs me about:

ok 1 testdatabase shouldn't print fail
ok 2 testrecord shouldn't print fail
ok 3 testsuminfo shouldn't print fail
ok 4 msibuild - invalid command line
ok 5 msibuild - empty MSI
ok 6 msibuild - UUID
ok 7 msibuild - UUID does not change
ok 8 msibuild - Set summary
ok 9 msibuild - add stream
ok 10 msibuild - add tables
ok 11 msibuild - add tables and definitions
ok 12 msibuild - seperate invocations
ok 13 msibuild - add table with data
not ok 14 msibuild - add table with stream
# (in test file
/mnt/MEDIA/portdev/pobj/msitools-0.106/msitools-v0.106/tests/tools.bats,
line 127) #   `output=$("$msiinfo" extract out.msi
Icon.firefox.16.0.2.0.ico.exe | sha1sum)' failed with status 127 #
/mnt/MEDIA/portdev/pobj/msitools-0.106/msitools-v0.106/tests/tools.bats:
line 127: sha1sum: command not found # rm -rf /tmp/tmp.XcCfGNiwDR ok 15
msibuild - update _SummaryInformation table ok 16 wixl - invalid
command line ok 17 wixl - WiX tutorial SampleFirst ok 18 wixl -
SampleUser ok 19 wixl - SampleMachine
not ok 20 wixl - stable component GUIDs
# (in test file
/mnt/MEDIA/portdev/pobj/msitools-0.106/msitools-v0.106/tests/wixl.bats,
line 55) #   `[ "$out" = "$exp" ]' failed # rm -rf /tmp/tmp.XHWv2Uuxoj
ok 21 wixl - WiX tutorial SampleFragment
ok 22 wixl - preprocessor variables
ok 23 wixl - preprocessor include & condition
ok 24 wixl - system include directory
ok 25 wixl - ARP example
ok 26 wixl - Binary/CustomAction
ok 27 wixl-heat - simple
ok 28 wixl - XML error
ok 29 wixl - UI ext

Seems it assumes gnu-style sha1sum binary naming rather than our sha1(1)

I can try and contribute a patch upstream fixing this portability issue
later. Thanks for catching that.

-- 
iz (she/her)

> I say mundane things
> so the uninteresting
> just might get noticed.

izder456 (dot) neocities (dot) org