save() and html_report() prevent further metric gathering

[nedbat]

Originally reported by Michael Crouch (Bitbucket: creidieki, GitHub: creidieki)

---

When you call coverage.save() it seems to prevent anything else from being read in, for example, html reports. For example, the following program shows 2 lines covered (the bodies of f1 and f2):

#!python

import coverage

cov = coverage.Coverage()
cov.start()

def f1():
    print 2

def f2():
    print 3

f1()
#cov.save()
f2()
cov.html_report()

However, if you uncomment the "cov.save()" line, it only shows the body of f1() as covered -- it doesn't record the body of f2().

cov.html_report() has the same problem when it's run between f1() and f2().

I'm not sure if this behavior is intended, but the documentation of save() and html_report() don't mention it.

I encountered this behavior while trying to find a "quick and dirty" solution for finding coverage information for the server side of a client/server application. I added cov.start() to the server initialization and added cov.html_report() to an existing server call, with the plan on just making that call to write out the currently-collected coverage information.

---

• Bitbucket: https://bitbucket.org/ned/coveragepy/issue/448

[nedbat]

Original comment by Matthew Boehm (Bitbucket: mattboehm, GitHub: mattboehm)

---

Thanks for reporting this, I agree that this is confusing and needs to be fixed.

Looking at the code, it appears that many of these functions (report, html_report, xml_report, save, combine, annotate) end up calling get_data, which resets the collector. While this side effect is documented in get_data, it's not documented in any of these other functions.

As I see it, there are 3 options:

1. Update the documentation for all these functions to say that they reset the collector
2. Add a reset_collector=True parameter to these functions
3. Update the functions so that they don't reset the collector

My guess is that option 3 would break other code out there, but I'll let others weigh in on this.

[nedbat]

Original comment by Loic Dachary (Bitbucket: dachary, GitHub: dachary)

---

What about implicitly calling close() when html_report() & other similar functions are called ? I have a feeling this is not the right thing to do but I'm not sure. What do you think ?

https://bitbucket.org/ned/coveragepy/pull-requests/106/implicitly-close-on-html_report-etc/diff

[nedbat]

Original comment by Loic Dachary (Bitbucket: dachary, GitHub: dachary)

---

It has something in common with save() prevents harvesting on stop()?

[nedbat]

Fixed in 12ea6d45d8b8 (bb)

[nedbat]

This is in 4.4b1.