lua-users home
lua-l archive

[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index] [Thread Index]


On 27 November 2015 at 11:23, Nagaev Boris <bnagaev@gmail.com> wrote:
> On Thu, Nov 26, 2015 at 3:56 PM, Roberto Ierusalimschy
> <roberto@inf.puc-rio.br> wrote:
>> The "basic tests" (with _U=1) are far from minimal. They already cover a
>> lot of ground (I would guess its converage is >90%),
>
> Is test coverage measured with some tool? Is it lines-of-code coverage
> or branch coverage? I would like to read 10% (or 1%) of non-covered
> code (and maybe cover it or find bugs in it). Lua's code is too small
> not to be covered 100.0%.

You can measure it yourself with e.g. gcov.
The bits that aren't covered by the tests are mostly handling out of
memory errors in odd places and platform specific code.

> I do not understand this point. More testers => more errors found. If
> tests are too complicated, maybe it is time to reorganize them using
> Busted [1].

The "full" tests are hard to run because they require recompiling lua
with certain flags.

busted is for testing lua code; it would be useless at testing lua itself.