989 bytes, patch
|Details | Diff | Splinter Review|
From Bugzilla Helper: User-Agent: Mozilla/4.75 [en] (Windows NT 5.0; U) BuildID: Create absolutely positioned layer. Insert into it a table whose width has been set with a percent value, E.G. 100%. The layer collapses down to its smallest wrappable size instead of following the table. This does not occur if the table's width is set using pixels/non-percents or if the layer uses relative positioning. Reproducible: Always Steps to Reproduce: 1. create an absolutely positioned layer 2. insert a table into the layer 3. set the tables width using a percentage value Actual Results: The layer is collapsed down to smallest width the table can assume. Expected Results: It should render the layer to be same width as table
Confirming the bug with Win98 20001007. Renaming the summary, as this bug has *nothing* to do with layers.
Sorry. I was using "layer" in the general sense to describe a CSS-P element. Habit I've got into from working with DHTML in Netscape 4 and IE4+. This bug has crept in recently, as it used to be OK in PR1. It developed some time after that...
This is still a problem on Mozilla Build 2001012205 Win32.
QA contact update
This bug is still lurking about and seems to have been forgotten. It is causing a lot of problems with a few sites since it now seems to have changed behaviour to be even worse. Up to now the content of a table whose width was set using percentages collapsed down to the length of the available text wrap. Now it is collapsing down to its smallest available state. Tables in Abs Pos DIV tags is a very common layout technique and this bug makes it unusable as a design technique. For an example of the problem, look at www.diuturnal.com.
The bug is that the green <div> in the testcase given is getting sized to 0, when it should be getting sized just wide enough to contain the text. Note that the green box should NOT span the width of the viewport as the testcase asserts. amar, could you create a valid minimal testcase for this bug?
This IS a valid test case. Either that or Netscape 4 IE4,5,6 and Mozilla up until about M8 are all wrong.
Why should the green box not span the width of the viewport?
Created attachment 41772 [details] [diff] [review] untested patch ( e.g. regression test still needed)
The patch has passed the regression test, the code was already there, we asked for the width of the reflow state but did not use it, e.g. bypassed the column width balancing.
fix checked in
Bernd, <div style="position:absolute> does not seem to work properly. Its not extending the border to the width of the text.. instead its sizing to 0.. <div style="position:relative> is working good..
which build you are using? wfm win98 2001071704 trunk (I dont care about branch)
Bernd, I was using Branch build when I reported my previous message.. I tested with trunk build its working fine.. win2000 Build ID# 2001-07-18-10-trunk win98 Build ID# 2001-07-18-10-trunk Marking verified