Don't base pass-otp availability decision on hardcoded /usr/lib

[nh2]

/usr/lib does not exist in various Linux distributions, nor does it work if you have installed your qtpass to a custom prefix such as /usr/local or into your home directory.

Disable this check for now, so that the OTP functionality can be enabled in such cases.

As a result, ConfigDialog::isPassOtpAvailable() now only has the functionality to return false on platforms where the OTP functionality is certainly not supported.

CC @FiloSpaTeam as the author of https://github.com/IJHack/QtPass/commit/5fc6c652af6b6cc949b571c090d5e6e65228b135 (via #407); also mentioning #394 and #327 so that this PR shows up there for others that, like me, searched for this problem.

---

A better future detection logic might be to run pass otp --version and check if that succeeds, but I'm not doing that here because it's more sophisticated and I'd like qtpass to work well with pass-otp for all users as soon as possible.

[coveralls]

Coverage increased (+0.02%) to 7.105% when pulling 2ca9f0ec5a8d709c97a2433c5cd814040c82d4f3 on nh2:patch-1 into b0cdc0098ad43d5ee08051ff83b52eaf77de7264 on IJHack:master.

[codecov[bot]]

Codecov Report

Merging #499 into master will increase coverage by <.01%. The diff coverage is 0%.

@@            Coverage Diff            @@
##           master    #499      +/-   ##
=========================================
+ Coverage    9.36%   9.37%   +<.01%     
=========================================
  Files          46      46              
  Lines        2925    2924       -1     
=========================================
  Hits          274     274              
+ Misses       2651    2650       -1

Impacted Files	Coverage Δ
src/configdialog.cpp	0% <0%> (ø)	:arrow_up:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update b0cdc00...2ca9f0e. Read the comment docs.

[maciejsszmigiero]

I think the pass OTP extension path should simply be made (compile-time) configurable, so distributions can set what path they install this extension at. OTOH, if you compile QtPass yourself you can set it as you like, even to /bin/true.

OTP module autodetection is useful, please don't remove this functionality.

[nh2]

I think the pass OTP extension path should simply be made (compile-time) configurable, so distributions can set what path they install this extension at.

That'd do as well.

I didn't do it here because it's more sophisticated and I won't have time to do it.

While I'd be happy with that, I also am not fully convinced that that's how pass envisioned it to be used; from man pass you can set PASSWORD_STORE_EXTENSIONS_DIR to anything you like at run time, thus also allowing user-provided extensions (for example in ~/.pass/.extensions) that are not available at compile time. That's why I think that checking e.g. pass-otp --version may be even better than your proposed /bin/true solution.

OTP module autodetection is useful, please don't remove this functionality.

Personally I think it's a good approach to make a quick fix available immediately and work on an improvement afterwards (especially for a convenience feature like force-disabling a check box), versus having it broken until the improved solution is available.

We must remember that the only thing this check does is to prevent the user to turn on an off-by-default checkbox by hand (or at least it seemed to be off-by-default on my system), which I'd consider not really that useful compared to not being able to use it at all.

[annejan]

Personally I think it's a good approach to make a quick fix available immediately and work on an improvement afterwards (especially for a convenience feature like force-disabling a check box), versus having it broken until the improved solution is available.

I agree with this . .