Quick request for folks reporting bugs...

[Joseph Kesselman <ke...@us.ibm.com>]

When you send an attachment to Bugzilla, it gives that attachment a
completely meaningless ,cgi filename. Please use the comments field to tell
us what to rename the file back to before using it. In particular, if it's
a .tar.gz file or other kind of archive other than a Windows-style zipfile,
please be explicit about that.

BTW, for anyone who isn't familariar with those:  to unpack them on
Windows, use
      ren whatever.cgi whatever.tar.gz          (gunzip insists on seeing a
.gz extension)
      gunzip whatever.tar.gz              (produces whatever.tar)
      tar -xvf whatever.tar               (produces however many files)
gzip/gunzip is widely available as a free download. I presume there's a
free implementation of tar somewhere, but I don't know where; I've been
using one of the commercial ports of that tool.


While I'm writing: Attachments/inclusions that demonstrate a problem --
preferably the smallest set of small files you can come up with that will
still provoke the breakage -- are *TREMENDOUSLY* helpful in bug reports.
Being able to actually watch the malfunction happen in a debugger makes
finding and fixing problems much easier. I know, sometimes it's extremely
difficult to reduce a huge testcase down to the key features that show the
bug, and sometimes you really do need a large file to provoke a problem...
but anything you can do to help us focus in on the specific cause of the
glitch will speed up our response to it, so it's in everyone's interest to
make the bug reports as specific and self-demonstrating as possible.

______________________________________
Joe Kesselman  / IBM Research

Re: Quick request for folks reporting bugs...

[Tom Amiro <To...@Sun.COM>]

I've run into this problem too with attachments. In some cases, it can 
be avoided by selecting the correct file type. It may seem a little odd,
but when you are attaching .xml or .xsl files, if you say they are 
plain text, the attachments come through OK.

Tom

Joseph Kesselman wrote:
> 
> When you send an attachment to Bugzilla, it gives that attachment a
> completely meaningless ,cgi filename. Please use the comments field to tell
> us what to rename the file back to before using it. In particular, if it's
> a .tar.gz file or other kind of archive other than a Windows-style zipfile,
> please be explicit about that.
> 
> BTW, for anyone who isn't familariar with those:  to unpack them on
> Windows, use
>       ren whatever.cgi whatever.tar.gz          (gunzip insists on seeing a
> .gz extension)
>       gunzip whatever.tar.gz              (produces whatever.tar)
>       tar -xvf whatever.tar               (produces however many files)
> gzip/gunzip is widely available as a free download. I presume there's a
> free implementation of tar somewhere, but I don't know where; I've been
> using one of the commercial ports of that tool.
> 
> While I'm writing: Attachments/inclusions that demonstrate a problem --
> preferably the smallest set of small files you can come up with that will
> still provoke the breakage -- are *TREMENDOUSLY* helpful in bug reports.
> Being able to actually watch the malfunction happen in a debugger makes
> finding and fixing problems much easier. I know, sometimes it's extremely
> difficult to reduce a huge testcase down to the key features that show the
> bug, and sometimes you really do need a large file to provoke a problem...
> but anything you can do to help us focus in on the specific cause of the
> glitch will speed up our response to it, so it's in everyone's interest to
> make the bug reports as specific and self-demonstrating as possible.
> 
> ______________________________________
> Joe Kesselman  / IBM Research

-- 
 Tom Amiro -- SQA Engineer
 Sun XML Technology Development
 voice: 781-442-0589 Fax: 781-442-1437
 eMail: tom.amiro@.sun.com

Exponential formats and XSLT format-number function?

[Rick Bullotta <ri...@lighthammer.com>]

The default output for something like Java's Double.toString() method may
format a value in exponential form, such as 1.57E-2.  However, it seems that
the format-number function does not handle this type of input (although
Double.parseDouble would).

Is this a bug or a designed behavior?  Is there a W3C or ISO standard for
string representation of floating point values in XML?

Thanks for any wisdom.

Rick Bullotta
CTO
Lighthammer Software (www.lighthammer.com)