Added progress report to console when running sourmash compute with --singleton

[lgautier]

Print a progress report when computing signatures.

• Is it mergeable?
• make test Did it pass the tests?
• make coverage Is the new code covered?
• Did it change the command-line interface? Only additions are allowed
  without a major version increment. Changing file formats also requires a
  major version number increment.
• Was a spellchecker run on the source code and documentation after
  changes were made?

[codecov-io]

Current coverage is 76.69% (diff: 100%)

Merging #84 into master will increase coverage by 0.10%

@@             master        #84   diff @@
==========================================
  Files            17         17          
  Lines          2273       2283    +10   
  Methods          46         46          
  Messages          0          0          
  Branches         84         84          
==========================================
+ Hits           1741       1751    +10   
  Misses          507        507          
  Partials         25         25          

Powered by Codecov. Last update 86adefe...3fb7871

[luizirber]

Hmm, I think we should avoid adding too many of these checks to the code. Is it too bad to just not flush the output?

And not for this PR, but we should definitely start using the logging module for this kind of output, instead of printing stuff. Also makes it easier to filter what output we want (INFO for printing progress, for example?)

[lgautier]

Hmm, I think we should avoid adding too many of these checks to the code. Is it too bad to just not flush the output?

It is a progress report. Not flushing the output kind of defeats the purpose.

And not for this PR, but we should definitely start using the logging module for this kind of output, instead of printing stuff. Also makes it easier to filter what output we want (INFO for printing progress, for example?)

Big yes about using logging (I am trying to match the current code base and not have changes and new features are the same time to the extent possible - I am happy to go both if not creating too much trouble). Note: Flushing is implicitly done by logging with each call, I believe.

[betatim]

Why not call sys.stderr.flush() in both cases?

[lgautier]

Would work... although it doesn't make the implicit point that Python 2 should eventually go ;-)

[ctb]

I like the lightweight functionality I'm using in syrah

…

--

[ctb]

If interested in logging, a super lightweight way to go about it is here:

https://github.com/dib-lab/syrah/blob/master/syrah#L25

Also see discussions/code here and here - the latter is probably better suited to sourmash, which is much lighter-weight than khmer.

(See #89)

[lgautier]

@luizirber : I only realize now that I missed the one little element (easy to miss) that might have caused the question about flush. I am using end="\r" in the print statement. This has the benefit of making the a progress prints stay on the same line in the terminal, but it is also requiring a flush to be sure that the output on the screen is updated (the character for a new line triggering a flush of stderr).

Python 3, you can try the following standalone snippet:

import time
for x in range(1, 101):
    print('iteration %i' % x, end="\r", flush=True)
    time.sleep(.5)

[ctb]

we have found little harm in keeping Python2.7 compat in khmer, and it (generally) takes little effort to maintain once established and under CI.

[ctb]

This needs to be updated for latest master. @lgautier I would have to create a new branch & PR to do so - is that OK?

[lgautier]

@ctb - sure. no pb. go ahead.