gistfile1.txt

Response to https://peerj.com/preprints/826v1/

1.

The method is dubious with respect to the conclusion.

The researchers do not analyse the code using any formal
or even established reliable methods. Instead, the metric that is used is "look
to see if some *detected* bug, which was fixed *post-release*.

Perhaps arguably even more dubious is the means by which "related to goto" has
been established. This is under-specified in the study and existing fields make
it clear that this is at least a non-trivial matter. 

This study:

* doesn't cover bugs that found and fixed pre-release.

* doesn't cover bugs that were not found at all.

* doesn't establish any relationship to "goto" that is recognised by existing
  research.

This is not to say the study is itself dubious. The conclusions that are drawn
from it are simply excessive; not by a little bit.

2.

If avoiding goto bugs was trivial or even guaranteed, as a non-total
construction it destroys the programmer's ability to apply
equational reasoning and parametricity.  Ergo, it is harmful.