[Users] See the 'FILES' section of `man claws-mail` for brief descriptions of the files.

[Dave Howorth]

On Thu, 8 Mar 2018 20:58:19 +0200
Removed GDPR <removed-gdpr at example.com> wrote:

> > Offer to pay that developer's customary consulting fee and you might
> > a more complete answer.  
> 
> Fair enough. How much do I have to pay and to who in particular so
> that I know which files do (not) need backup?

Full marks to Andrej for his recent intervention, and I too have
noticed that your questions are well-intentioned although your way of
asking sometimes appears thoughtless.

If we look at your question above,and with my perspective as a
developer although NOT a Claws developer, I would suggest you first need
to ask yourself some further questions and answer them.

Firstly, is it a serious question?
Secondly what level of accuracy and completeness would you want to pay
for?
And thirdly, what warranty would you expect?

It seems quite likely to me that nobody has ever thought about the
notion of an exact backup. It seems much more likely that developers'
concerns have centred on things like not crashing if there is an
inconsistency between the claws file collection and the various remote
message stores for any reason at all, of which an inconsistent backup
would be just one possible case. If they had thought about the notion,
I'd have expected them to step forward by now, especially in response
to your present question.

So answering your question is going to involve somebody spelunking the
code, trying to determine the result of various situations. And that
won't be easy, given that the code (probably) wasn't written with that
as an identified requirement. It will likely involve a lot of code
reading and a bunch of experiments, such as making a whole load more
tests for claws. And even after they've done that, they won't be sure
they've caught all the possible cases, and they won't want to be
committed forevermore to fixing and documenting whatever new use case
you come up with. They certainly won't want to be liable for the
possibility of any consequential damages.

Set against that we notice that nobody, neither a developer nor any
other user, has apparently ever had a problem by just taking a full
backup and restoring from it. Sure they may have had some local
breakage of some message stream or lost a few messages, but things like
that can usually be recovered out of band - by going to a web archive,
by asking senders to resend etc. So the community seems to have
historically judged that there's little value in answering your
question.

So you'd be asking a developer to work on something essentially of use
just to you, in preference to work that would benefit everybody. That's
not an attractive proposition. Further, there's the risk of scope
growth and ongoing support issues (depending on your answers to my
supplementary questions). So anybody standing up to offer to answer
your question is going to want to be well paid for their time on what's
likely to prove a tedious and thankless task. Are you up for spending
thousands of dollars to answer your question?

An alternative possibility might actually be for you to answer the
question yourself. If you were prepared to write tests and/or learn to
read the code in order to answer your question, and then write it up
for the documentation, I'm sure the developers would be more than happy
to help you learn how to do it. It seems to me that you have the
necessary attitude and abilities. It's not rocket science.