On Sun, 2002-05-05 at 01:54, Rainer Klute wrote:
> People,
> 
> I neither have the time nor the interest for taking part in a 
> lengthy discussion on little-endians. Here's the HPFS point 
> man's (e.g. mine) point of view:
> 
> - Andy posted a bug that there should be only a single 
>   little-endian package throughout POI. I accepted that bug (see 
>   <http://nagoya.apache.org/bugzilla/show_bug.cgi?id=8354>) as 
>   being a very valid point.
> 

That wasn't the beginning of this, but okay.

> - I also said that I'd look at it when time permits and that 
>   this bug is not an urgent one. So far no one disagreed. The 
>   conclusion is that there is no immediate need for action. That 
>   is, org.apache.poi.hpsf.littleendian will exist for quite a 
>   while and then disappear.
> 

Umm I now disagree.  Its been quite a while.  

> - However, a piece of software who's days are counted still 
>   needs to be maintained if an error appears. This is what Drew 
>   pointed out and this is something I agree with.
> 

-1, its my right to object.  I object.  I've given my reasons and
alternative.

> - We can now pick one of three options:
> 
>   - Integrate both little-endian packages, pulling together an 
>     at least partly new API and make the rest of POI use it. I don't 
>     oversee what amount of work this would be but I don't think is 
>     can be done in an hour or two. Like Drew I also have my own 
>     little schedule and cannot do this before middle of June.
> 
>     If anyone else has the time to do it: go for it.
> 

I don't have time.  But I'm making time.  This is precisely what I'm
doing.  However I have to do this in pieces.  My first step will be to
remove the old so that I get compilation errors and my second step will
be to fix them.  If you don't like my method you're free to do it
*another way*.  

>   - Maintain org.apache.poi.hpsf.littleendian for the rest of 
>     its lifetime. Maintaining means fixing bug but not adding 
>     functionality (here: let DWord.intValue interpret the bytes 
>     as an unsigned int and return that). Comments are not part 
>     of functionality but could be misleading. Anyway, it does 
>     not hurt to correct them.
> 

-1, the cancer has grown.  Its time to cut it out.

>   - Leave it as it is for the time being. Given the usually 
>     small amounts of data in property set stream I guess the 
>     probability for the signed vs. unsigned bug is relatively 
>     unlikely to appear.
> 
> These are the facts as I see them. You might see them 
> differently. However, a different point of view does not justify 
> any harsh wordings in e-mails. Showing the Big Hammer, threatening 
> to delete a package or saying other harsh words is only frustrating 
> and demotivating.
> 

Sorry, maybe my colorful speech covered the simple facts:

1. We gave you time.
2. Its time to fix this.
3. I said if you don't I will.
4. I am now.
5. If you don't like how I'm going to do it, well your course of action
is to make time and do it yourself.
6. De-motivating and frustrating you isn't my goal.  Its fixing an
important code quality issue.

We can take a vote however if you read, +1 = "I agree and will put
forward the effort", so basically its put up or live with how *I* do it.

Sorry it has to be this way,

Andy

> Best regards
> Rainer Klute
> 
>                            Rainer Klute IT-Consulting GmbH i. Gr.
>   Dipl.-Inform.
>   Rainer Klute             E-Mail:  [EMAIL PROTECTED]
>   K�rner Grund 24          Telefon: +49 172 2324824
> D-44143 Dortmund           Telefax: +49 231 5349423
> 
> 
> 
-- 
http://www.superlinksoftware.com - software solutions for business
http://jakarta.apache.org/poi - Excel/Word/OLE 2 Compound Document in
Java                            
http://krysalis.sourceforge.net/centipede - the best build/project
structure
                    a guy/gal could have! - Make Ant simple on complex Projects!
The avalanche has already started. It is too late for the pebbles to
vote.
-Ambassador Kosh

Reply via email to