libnml: Fixed memleak in configuration parsing

[smoe]

Also suppressing false positive report on use of undefined variable.

[BsAtHome]

Instead of suppressing the cppcheck uninitialized warning, it may be appropriate just to initialize the pl variable like:

double p1[9], p2[9], pl[9] = {};

No comment on the mem-leak :-)

[smoe]

Instead of suppressing the cppcheck uninitialized warning, it may be appropriate just to initialize the pl variable like:

double p1[9], p2[9], pl[9] = {};

I would like to keep the suppression as a reminder that the code is a bit tricky to read. Also, erroneously assigning a default value 0 or undefined is equally unfortunate, so the suppression is something like a quick statement that the devs are aware of the situation.

No comment on the mem-leak :-)

:)

[BsAtHome]

I would like to keep the suppression as a reminder that the code is a bit tricky to read.

Tricky,... that is an understatement.

Also, erroneously assigning a default value 0 or undefined is equally unfortunate

That is not necessarily a problem. The real problem is that cppcheck cannot perform the correct analysis. The code and compound condition makes first and the value of pl mutually exclusive.

The cppcheck message is a false positive. If you suppress the message, then you should at least precede it with a comment why the message needs to be suppressed. That would help to understand the "tricky" code a bit better ;-)

[smoe]

Here the explanation is given why the cppcheck error is suppressed.

[BsAtHome]

I'm sorry to be a nitpick here...

The comment is usually added before the suppress with a reference to the suppress.

// Suppress cppcheck false positive: 'first' == 1 when 'pl' is undefined and therefore not a problem
// cppcheck-suppress uninitvar

[smoe]

@BsAtHome, many thanks, it is indeed much nicer with the comment preceding, done. Broke the comment into two lines since it became a bit long (108 characters). @andypugh - ready?