Handling Batik bugs

[Peter Becker <pb...@meganesia.int.gu.edu.au>]

Thomas E Deweese wrote:
> 
> >>>>> "PB" == Peter Becker <pb...@meganesia.int.gu.edu.au> writes:
> 
> PB> Thomas E Deweese wrote:
> >> This bug should be fixed in the current CVS of Batik.
> 
> >>>>> "PB" == Peter Becker <pb...@meganesia.int.gu.edu.au> writes:
> 
> PB> Currently I distribute my two patches (this one and the white
> PB> edges problem) as class files with my program -- is this ok or
> PB> should I use the CVS version? Will there be an 1.1 or 1.0.1
> PB> release in the next time?
> 
>     Well, that decision is up to you.  I will point out that I believe
> your publicly announced fix for the white edge problem, of subtracting
> .5 instead of rounding is probably wrong, the one I committed does
> nothing with .5 and just calls Math.round (originally the code added
> .5 and then truncated to int, but this gives bad answers for negative
> numbers, I then moved to Math.round, but missed to removing the .5
> rounding factor).

This fix was just a quick hack -- worked for the few examples I tried
but since I didn't really understand the code I didn't even try to get a
real fix. I'll switch to Math.round, too.

The problem I have with switching to the CVS version is that CVS
versions tend to get all the new errors ;-) Or do you have a bugfix
branch? I couldn't find one.

[...]

> PB> Is there a list/place to stay up to date with Batik problems
> PB> without too much work? Bugzilla is quite complex and noisy, an
> PB> announcement list would be cool ;-)
> 
>     Well, there is batik-dev which probably isn't much better, or
> xml-batik-cvs which would allow you to see commit messages (not 100%
> certain if that is the public name for it).  I'm not sure an
> announcement list would cover this, since I don't think I would
> necessarily post such a bug fix to an announcement list (which I
> normally associate with 'big' announcements like version 1.0
> released).

I guess a bugfix branch would be perfect for me -- cvs logs are a quite
nice way of communicating since they give much context. I just would
like to know about problems in my libraries before I try to fix them in
my own code ;-)

Don't get me wrong: I think it is quite unavoidable that libraries have
bugs and I understand that administrating bugs is not your favoured job
;-) But it is very inconvenient to try to fix a bug in your program for
hours and then to realize that it is a well known (and maybe already
fixed) problem in a library -- didn't happen with Batik yet, but it is
not that uncommon.

[...]

Regards,
  PeterB

---------------------------------------------------------------------
To unsubscribe, e-mail: batik-users-unsubscribe@xml.apache.org
For additional commands, e-mail: batik-users-help@xml.apache.org

Re: Handling Batik bugs

[Vincent Hardy <vi...@sun.com>]

Peter,

We have not created any branch yet. May be we should create a CVS
log of all the bug fixes and document this in the weekly builds. 
There have been no weekly builds recently because of conferences and 
some other issues (personnal unavailabilities), but we will do our 
best to start again soon.

Vincent.

Peter Becker wrote:
> 
> Thomas E Deweese wrote:
> >
> > >>>>> "PB" == Peter Becker <pb...@meganesia.int.gu.edu.au> writes:
> >
> > PB> Thomas E Deweese wrote:
> > >> This bug should be fixed in the current CVS of Batik.
> >
> > >>>>> "PB" == Peter Becker <pb...@meganesia.int.gu.edu.au> writes:
> >
> > PB> Currently I distribute my two patches (this one and the white
> > PB> edges problem) as class files with my program -- is this ok or
> > PB> should I use the CVS version? Will there be an 1.1 or 1.0.1
> > PB> release in the next time?
> >
> >     Well, that decision is up to you.  I will point out that I believe
> > your publicly announced fix for the white edge problem, of subtracting
> > .5 instead of rounding is probably wrong, the one I committed does
> > nothing with .5 and just calls Math.round (originally the code added
> > .5 and then truncated to int, but this gives bad answers for negative
> > numbers, I then moved to Math.round, but missed to removing the .5
> > rounding factor).
> 
> This fix was just a quick hack -- worked for the few examples I tried
> but since I didn't really understand the code I didn't even try to get a
> real fix. I'll switch to Math.round, too.
> 
> The problem I have with switching to the CVS version is that CVS
> versions tend to get all the new errors ;-) Or do you have a bugfix
> branch? I couldn't find one.
> 
> [...]
> 
> > PB> Is there a list/place to stay up to date with Batik problems
> > PB> without too much work? Bugzilla is quite complex and noisy, an
> > PB> announcement list would be cool ;-)
> >
> >     Well, there is batik-dev which probably isn't much better, or
> > xml-batik-cvs which would allow you to see commit messages (not 100%
> > certain if that is the public name for it).  I'm not sure an
> > announcement list would cover this, since I don't think I would
> > necessarily post such a bug fix to an announcement list (which I
> > normally associate with 'big' announcements like version 1.0
> > released).
> 
> I guess a bugfix branch would be perfect for me -- cvs logs are a quite
> nice way of communicating since they give much context. I just would
> like to know about problems in my libraries before I try to fix them in
> my own code ;-)
> 
> Don't get me wrong: I think it is quite unavoidable that libraries have
> bugs and I understand that administrating bugs is not your favoured job
> ;-) But it is very inconvenient to try to fix a bug in your program for
> hours and then to realize that it is a well known (and maybe already
> fixed) problem in a library -- didn't happen with Batik yet, but it is
> not that uncommon.
> 
> [...]
> 
> Regards,
>   PeterB
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: batik-users-unsubscribe@xml.apache.org
> For additional commands, e-mail: batik-users-help@xml.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: batik-users-unsubscribe@xml.apache.org
For additional commands, e-mail: batik-users-help@xml.apache.org