Re: [PATCH] git_blame2: fix rendering on Firefox 3.0b5

[Jakub Narebski <jnareb@gmail.com>]

On Tue, 22 April 2008, Steven Walter wrote:
> On Tue, Apr 22, 2008 at 1:05 PM, Jakub Narebski <jnareb@gmail.com> wrote:
>> On Tue, 22 April 2008, stevenrwalter@gmail.com wrote:
>>>
>>> Using variably-sized rows and rowspan does not render correctly with
>>> Firefox 3.0b5.  Instead, make each row have the same number of cells.
>>
>>  Beta version of Firefox doesn't correctly render perfectly valid HTML;
>>  more, the way rowspan was meanto to be used, see example in
>>
>>   http://www.w3.org/TR/html401/struct/tables.html#h-11.2.6.1
>>   (subsection "Cells that span several rows or columns" in HTML 4.01)
>>
>>  and you want gitweb to generate invalid HTML?
> 
> What's invalid about the way my patch generated HTML?  It renders
> correctly on both FF2 and FF3.

The fact that HTML is valid or not is not dependent on the fact that
some web browser, or some beta version of web browser (which can have
bugs), renders it correctly or not.  You check if HTML is valid against
HTML specification, e.g. using one of online HTML validators, including
W3C Validator or WDG Validator.

> It may have been fully correct before, but is it any less correct now?

The idea is that enlarged cell (either via rowspan, or via colspan),
takes place / pushes out other cells.  Se the ASCII-art diagram (to be
viewed using fixed-width, non-proportinal font) below:

   |-----|-----|
   |  A  |  B  |   <tr><td rowspan="2"> A </td><td> B </td></tr>
   |     |-----|
   |     |  C  |   <tr><!--   space for A   --><td> C </td></tr>
   |-----|-----|
  
See example in provided link to HTML specification.  It is natural way
of writing HTML table with cell spanning multiple rows.

> Conversely, I will also file a bug with Firefox regarding it's
> apparently-broken handling of rowspan.

If handling of rowspan differs from handling colspan with respect to
variable number of cells in row / column, then it is certainly a bug.
Gitweb don't use colspan beside row spanning whole table, so you
wouldn't detect it.


All said, the fact that beta version of a browser doesn't render
gitweb correctly, while HTML is correct and follows examples in
HTML specification, is not a reason to change gitweb: it is reason
to file bug report, and try newer version.

If some popular browser had trouble rendering some complicated HTML
code, we could think of changing gitweb.

P.S. I have dug your original letter from spam folder on my GMail
account, and as far as I can see (reading via GMane NNTP / news
interface) it didn't appear on git mailing list either, probably
because of VGER antispam filter.

P.P.S. All that aside your commit message needs improvements.
First line (summary line) should IMHO start with "gitweb: ",
not "git_blame2", and tell _what_ was fixed.
-- 
Jakub Narebski
Poland