Red Hat Bugzilla – Bug 53925
keyboard focus in scrollbar loses input
Last modified: 2007-04-18 12:37:17 EDT
Using the default RedHat setup, start an XTerm with the Scrollbar visible. For example, "xterm -sb". Now give the xterm focus, and note that your typing is echoed. Now, place the mouse pointer in the scrollbar and note that your typing is not echoed. To Fix: Edit the file /usr/X11R6/lib/X11/app-defaults/XTerm. Towards the end of this file, change the line *VT100*translations: #override \ to *VT100.translations: #override \ (note the change from asterisk to period.) Start up a new XTerm, and note that typing in the scrollbar is now correctly echoed. The particular problem is due to a misdesign in X11, characterized by jwz in the Xscreensaver code: (xscreensaver.c:77) * The reason that we can't select KeyPresses on windows that don't have * them already is that, when dispatching a KeyPress event, X finds the * lowest (leafmost) window in the hierarchy on which *any* client selects * for KeyPress, and sends the event to that window. So, if the #override applies to everything below VT100 in the window hierarchy (which is what happens with the current RedHat setup, which uses VT100*translations) then the keypresses are delivered to the scrollbar widget, and do not arrive at the VT100 widget. Whereas if the #override applies just to the VT100 widget (which is what the change I propose causes to happen) then keypresses are delivered and interpreted correctly.
Problem confirmed in RHL 7.0, 7.1. Proposed solution confirmed to fix this behaviour also. Investigating if there are any potential negative side effects of making the change. If not, I'll add it to a future release.
I'm not sure what ramifications this sort of change would have. Past changes to xterm resources have been grave mistakes because they have fixed the problem reported by one user at the expense of breaking something else for another user. Also, we provide, but do not officially support xterm. Our supported terminals are konsole, and gnome-terminal. Although I wont make this change at this time, I am defering your request for future consideration.
If RedHat does not support xterm, then you should ship the vendor version. The configuration file you ship is NOT the version that's shipped with XFree86, and your changes BREAK the application. If you don't understand the resource syntax well enough to understand whether this bugfix will have other undesirable effects, perhaps you should forward the change to someone who does understand. Thomas Dickey (xterm maintainer) said > Regarding your analysis - that's how I read the code as well. But in > either case, making changes to app-defaults files should be done sparingly > to avoid breaking things that the user has already setup: once you have > set a pattern, the user can't set that pattern. The code in question IS a change to the app-default file, and the breakage is in the code that RedHat adds. If you can't fix bugs in the things YOU are adding to standard UNIX utilities, what kind of distribution are you?
>The code in question IS a change to the app-default file, and the breakage is in >the code that RedHat adds. If you can't fix bugs in the things YOU are adding >to standard UNIX utilities, what kind of distribution are you? We're evidently the kind of distribution that draws rude obnoxious users who feel the need to insult us. For a second there, I thought I was reading slashdot commentary and not bugzilla. It is bug reports like this one, that make me want to close all requests for enhancement on things like xterm and other similar stuff. We make these changes at our users requests usually - to show that we are open to our users needs and suggestions. It is not EVER possible to fully realize all of the implications a particular change may induce, especially if one is not intimately familiar with a proposed change. As such, if a given change is made, and makes it through the plethora of our own testing, plus the all of our public beta testing, without any bug reports streaming in - we can only conclude that a given change is functional. And for the record, I had absolutely nothing to do with any of these changes to xterm which have broke whatever they have broke for whoever they've broke them for. Nonetheless, I am the one who ends up now being negatively attacked. I don't know a heck of a lot about xterm at all, and am hesitant to make changes to it, in particular due to receiving negative bug reports like this one. I've nonetheless made the change of * to . in my current package, which will appear in rawhide. Aside from fixing "bugs that those idiots at Red Hat have stupidly inflicted on xterm users", I don't plan on ever making any future enhancements to xterm. It's much more productive for me to pass bug reports upstream and let them sort it out, and spend my time on areas of XFree86 that I am personally much more familiar with, and actually care about.