[bt-devel] Sword++ Warnings

[Jaak Ristioja]

Dear Gary,

I am sorry about the confusion caused through my fault. I could probably
have considered my words more carefully, as well as taken some more time
to think through and formulate the expectations more specifically and
with a bit more care. I was certainly not prepared to handle interest in
contributing to Sword++ so soon, and perhaps some of what I've said is
contradictory. As I wrote in my previous e-mail, I'm a better developer
than a project lead. Even worse as a spokesperson. I'm sorry for the
setback I've caused. I agree that I'm sometimes a difficult person to
work with.

One of the goals is to fix bugs. The changes just hid the problem even
more by silencing the warnings. Yes, you did add a run time error
message, but that doesn't win us much. While I agree, that as
developers, having zero warnings is a good thing, this should not come
by hiding warnings but by fixing or working around the fundamental
issues behind these warnings. Yes, fixing such would have been a huge
undertaking, so I see that suggesting to fix the warnings first was (at
least in part) a mistake. I'm still a bit at a loss about what else to
recommend for starters.

Yes, my changes to listkey.cpp introduced more warnings. I'll get to
those later. I think it is inevitable we get them in these early stages
of development. These are caused by wrong types being used. I corrected
some types to get rid of some signed-vs-unsigned and similar warnings,
only for several others of the same type popping up because of my fix.
Not that my fix was a step in the wrong direction, but that wrong types
are being used in much of the code. Given such huge problems with the
code, I could have written one mega-commit fixing the problem at many
places, but I decided to address the issue more incrementally with
several commits.

Jaak

On 02.10.2016 06:28, Gary Holmlund wrote:
> Jaak,
> 
> I see my changes to eliminate some warnings were removed. I had to
> compile the code, observer the warning came back, and read the git logs
> to find out about this.
> 
> In my solution I logged the fact that an error occurred primarily for
> developers benefit. This is in code that is very unlikely to have errors
> and has never had any error logic. I thought that you would not like the
> large code changes needed to completely test for errors in this code.
> 
> The code I had was certainly an improvement. I had the warnings using
> the GCC compiler down to zero. An important coding rule is that a
> project should compile at high warning level without warnings. Why?
> Because you want to be able to see if new warning are being introduced
> into the code. I have found that you have introduced several new warning
> in the code since my code was removed. See your changes in listkey.cpp
> for example.
> 
> It is hard to work on a project when the expectations are not very clear.
> 
> Gary
> 
> 
> _______________________________________________
> bt-devel mailing list
> bt-devel at crosswire.org
> http://www.crosswire.org/mailman/listinfo/bt-devel