#220 closed defect (fixed)
Selection hides other features in IE6
Reported by: | chrisclaydon | Owned by: | madair |
---|---|---|---|
Priority: | P1 | Milestone: | 2.0 |
Component: | Widgets | Version: | SVN |
Severity: | Major | Keywords: | |
Cc: | Browser: | IE6 | |
External ID: | 1193382.01 | Operating System: | All |
state: | Analysis Requested |
Description
This occurs in the fusion-mg21 branch:
Open a layout based on the Sheboygan sample map in IE6.
Select a feature.
The selection appears, but the rest of the map becomes obscured - this could be a problem with PNG transparency in IE6(?)
If a rectangle is dragged to select. An error dialog is displayed referencing an MgClassNotFoundException.
Change History (11)
comment:1 by , 16 years ago
Browser: | All → IE6 |
---|---|
state: | New → Analysis Requested |
comment:2 by , 16 years ago
Status: | new → assigned |
---|
comment:3 by , 16 years ago
Using the Fusion code from the fusion2-mg21 branch, I loaded up the slate template, zoomed into the Sheboygan Parcels layer, and selected a parcel. The selection overlay showed up with an opaque background that hid all the map data underneath. I tested with a real version of IE6 (6.0.2900.2180.xpsp_sp2_rtm.040803-2158) in a VM. It could be that IETester does not emulate the PNG transparency issues present in IE6.
comment:4 by , 16 years ago
I just downloaded IETester, and I can reproduce the problem with it too. Which operating system are you using, Mike? I have XP.
comment:5 by , 16 years ago
Also which version of MapGuide? I wonder if your deployment is reverting to the old selection mechanism (using GetMapImage)...
comment:6 by , 16 years ago
comment:7 by , 16 years ago
just re-reading RFC38 and found this:
We will rely on gd to perform the colorspace transformation for PNG8. Due to a limitation in gd, selection color transparency color will not be maintained. This means a PNG8 selection image will completely block out the features and labels underneath it.
Is this part of the issue?
comment:8 by , 16 years ago
Mike, I think that a simple check using Fiddler could tell you that the image coming back is PNG24.
On another note, why does it take so much work to get this reproduced? Something like this should be simple to diagnose. It just seems silly to me.
comment:9 by , 16 years ago
comment:10 by , 16 years ago
Resolution: | → fixed |
---|---|
Status: | assigned → closed |
comment:11 by , 16 years ago
As noted above, the issue was having an instance of MG to test against. MG2010 was restored and fix applied
I'm unable to reproduce any of these errors. Testing with IETester in IE6 mode, Fusion trunk and the fusion2-mg21 branch, standard template with the Sheboygan sample dataset (and Fusiont trunk with the slate template).