test whether postgres user exists

[ltalirz]

fix #18

So far, pgsu only tried the "sudo route" when it detected that it was
running on Ubuntu Linux.
It was reported that this test can fail when running on Windows
Subsystem for Linux.
Furthermore, this test might be too strict and exclude other Linux
distributions with similar postgresql setup (unverified).

Here, we switch from the check of the distribution name to a check
whether the postgres system user exists.

Furthermore, we introdyce the try_sudo and sudo_user constructor
arguments, which can be used to force pgsu to try the "sudo route" for
a given system user.

[ltalirz]

@zhubonan does this fix things for you?

[codecov-io]

Codecov Report

Merging #25 (17dbf86) into master (f453385) will increase coverage by 0.12%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master      #25      +/-   ##
==========================================
+ Coverage   95.70%   95.83%   +0.12%     
==========================================
  Files           2        2              
  Lines         163      168       +5     
==========================================
+ Hits          156      161       +5     
  Misses          7        7              

Impacted Files	Coverage Δ
pgsu/__init__.py	95.51% <100.00%> (+0.14%)	⬆️

---

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 f453385...17dbf86. Read the comment docs.

[ltalirz]

@zhubonan @greschd Does this make sense?

From the CI point of view it works just as well, but of course I'm not testing other distros here.
Let me know if you would like me to add any other platforms to the CI.

[greschd]

Yeah I think the code makes sense. I was a bit confused by the naming though: Why is it called unix_user? Shouldn't it be called something like postgres_user?

I think the reason this seems weird to me is that there are lots of "unix users" on any given system (or.. well, none), but only one of them is the postgres user.

[ltalirz]

Yeah I think the code makes sense. I was a bit confused by the naming though: Why is it called unix_user? Shouldn't it be called something like postgres_user?

I think the reason this seems weird to me is that there are lots of "unix users" on any given system (or.. well, none), but only one of them is the postgres user.

Yeah... the potential problem with postgres_user is that we also want to avoid confusion with the PostgreSQL database (super)user (which is also called postgres in most cases, even if there is no corresponding unix user).

I had sudo_user in some commits before but then also this was potentially confusing (user that is running sudo vs user that you change to)...

I guess postgres_unix_user or postgres_system_user would be the most ambiguity-free?

[greschd]

Yeah... the potential problem with postgres_user is that we also want to avoid confusion with the PostgreSQL database (super)user (which is also called postgres in most cases, even if there is no corresponding unix user).

Aah, yeah that makes sense.

I guess postgres_unix_user or postres_system_user would be the most ambiguity-free?

Yeah, I think that'd make sense. The choice between postgres_unix_user and postgres_system_user comes down to whether you may want to use the same variable on (bare, non-WSL) Windows at some point, I guess.

[zhubonan]

Makes sense for me.
I have tested and it works well for me on WSL1 Ubuntu 18.04 Windows 10.

[ltalirz]

pgsu 0.2.0 out on pypi

[greschd]

Thanks @ltalirz! Do you think we should relax the version constraint (currently ~=0.1.0) in aiida-core to ~=0.1, such that it will install 0.2?

[ltalirz]

oops, indeed!