Update function declarations to make docs reproducible

[lamby]

Whilst working on the Reproducible Builds effort, we noticed that satpy's documentation could not be built reproducibly. This is because it uses the absolute buildpath in the generated docs.

This was first reported in Debian as bug #919566.

[mraspaud]

Hi, thanks for your contribution.
IMHO, this reduces readability though, would you mind justifying your approach please ?

[coveralls]

Coverage decreased (-0.009%) to 77.424% when pulling 92a9271 on lamby:reproducible-build into 417b29f on pytroll:master.

[coveralls]

Coverage decreased (-0.009%) to 77.424% when pulling 92a9271 on lamby:reproducible-build into 417b29f on pytroll:master.

[codecov]

Codecov Report

Merging #579 into master will increase coverage by 0.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #579      +/-   ##
==========================================
+ Coverage   77.43%   77.44%   +0.01%     
==========================================
  Files         136      136              
  Lines       19143    19153      +10     
==========================================
+ Hits        14823    14833      +10     
  Misses       4320     4320

Impacted Files	Coverage Δ
satpy/composites/__init__.py	64.16% <100%> (+0.1%)	⬆️
satpy/scene.py	84.13% <100%> (+0.06%)	⬆️
satpy/readers/__init__.py	94.46% <100%> (+0.07%)	⬆️
satpy/config.py	47.82% <100%> (+1.15%)	⬆️

---

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 417b29f...92a9271. Read the comment docs.

[lamby]

@mraspaud

would you mind justifying your approach please ?

Not entirely sure what you mean by this, sorry. (Or, what do you not find readable about this?)

[mraspaud]

Sorry I wasn't clear: I meant why does the current implementation break the reproducibility of the documentation ?

About readability, I just find that turning eg this:

def get_environ_config_dir(default=PACKAGE_CONFIG_PATH):

into this:

def get_environ_config_dir(default=None):
    if default is None:
	    default = PACKAGE_CONFIG_PATH

hurts readability and documentation a bit: default is not None, it is an actual path.

But maybe that's just me :)

[djhoese]

While I agree with @mraspaud that the code becomes a little uglier, I think the benefits outweigh the ugliness. It moves the config directory determination to runtime rather than "compile time".

Additionally, this is something (similar) to what I am going to need to do in the future when I start using the donfig package as SatPy's configuration backend. This way users could easily modify the "default config" location during execution and it be retrieved from the global configuration object inside functions/classes.

I am OK merging this.

[lamby]

Sorry I wasn't clear: I meant why does the current implementation break the reproducibility of the documentation ?

Because it will encode the current build path via that constant (eg. /home/lamby/foo on my machine, /home/mraspaud/bar on yours gets embedded into the generated HTML) and we thus different results → not reproducible. :)

[mraspaud]

Ok, I'm merging this then