[parted-devel] [PATCH 1/2] tests: factor utility functions into "library"

[Jim Meyering]

Hans de Goede wrote:
> Would it not be better to initialize write back to 0, and only set it to 1
> when we actually want to do writeback ?

Yes, definitely, and I planned to do that (though you're welcome to, too).
With the three just-added write_back=0 stmts, it's clearly better to
initialize it to zero and then to set it to 1 in the less common code paths.

However, that change belongs in a patch separate from the bug-fixing
one.

> Also since the code paths for using the ok gpt in case the primary
> or backup is corrupt, only warn without asking anything, how can
> one cause the table to be fixed up without making any changes now ?

I haven't looked just yet, but have a fundamental aversion to
the "gpt_read" function modifying the partition tables,
so my hope is to move that functionality into a new function,
even if it means changing parted's UI.
Might be hard to do that without changing the API though.

> This is a parted problem I think no a libparted problem, as parted is
> instand apply, there is no commit command afaik, I case we should add
> one so issues like one of the 2 tables being corrupt can be fixed
> by using the commit command.

Suggestions welcome, and patches even more so ;)