The question-mark problem is due to the fact that it is a "weak" character, i.e. it has no fixed direction associated with it, unlike a Roman letter or Hebrew letter. Weak character handling seems to be notoriously difficult in bidirectional text, since the computer can't guess to which text run you intend your weak character to belong (I'm sure Simon Montagu can offer far better commentary on the technical issues). ;-) My method of working around issues with weak characters is to use the LRM and RLM characters as necessary to add directional hints.

Unfortunately, I get the feeling that for many software developers, usability is at best an afterthought, and companies that are lucky enough to have a beltzner on staff to guide their engineers are, well, just that: lucky!

It is very easy to find the bad things in other open source projects, but Firefox itself is far from perfect. Ask any Linux user out there - the freezes of Firefox happen for every user even on newer computers, and thanks to compiz freeze to black&white effect it is even more noticeable than before. We really need to do something in order to give Linux users better usable browser than today. Please help making Firefox 3.5 better for Linux users!

couldn't agree more.

firefox is sucking in two type of machines

linux, and windows netbooks.

Mozilla need to diverse at least 20% man power to improve these two area,, its simply the most developing areas.

Try running Firefox on kde4/linux your frustration level will rise up quite a lot.

The latest version of Colonization broke savegames from the previous one also, annoyingly.

Try hitting Ctrl-Up in Pidgin; it's been supported for years. It can't use the up arrow for this, because then you wouldn't be able to navigate multi-line text. Suggestions for how to make this more discoverable welcome...

Try hitting Ctrl-Up in Pidgin. That's been there for years. Suggestions for how to make it more discoverable welcome; it can't be Up, because then you wouldn't be able to navigate multi-line messages.

Why doesn't your blog give me any feedback whatsoever that my post has been queued when I hit Post? Remember the commenteeeeeeeeeeeerrrrrrrrrrsss.....

Will: sorry about that; I think there's a caching issue. But it's not my server or blog installation :-(

I guess my issue is that I don't see the need for multi-line messages. But then I use IRC only, and not other chat systems! Perhaps if you press Up when at the very top of a message, you could get a status bar message or other notification "Press Ctrl-Up to recall previous text"?

Gerv

Bear in mind, mixing RTL and LTR text in a WYSIWYG environment is inherently a somewhat difficult problem. Not that this completely excuses all horrors, but getting it totally right would be hard. You do seem to have had more than a reasonable amount of trouble with it, though. That font thing for instance sounds really weird and frustrating.

Regarding cell colors in Calc, I've never had any trouble changing entire rows or columns at once, and (plain) text pasted in subsequently stay that color.

Regarding the saved games in BfW, an unfortunate consequence of the open-source development model (and, particularly, the release early release often approach), when it comes to games, is that saved games from one release will often not really be applicable to a later release, because stuff changes, and I don't mean the saved-game format. (Actually, Wesnoth has a WML-based save-game format, though using it is not the default for some reason, probably to do with disk space, though frankly the saved games don't take up enough space to make that a significant consideration.) Stuff changes. Rules change. Units change (sometimes quite significantly). Campaigns and scenarios change. Wesnoth doesn't even necessary ship the same campaigns with different releases, and campaigns change internally. Entire scenarios disappeared from the Heir to the Throne campaign a couple of releases back (circa 1.2 IIRC). You know that weird inexplicable statement in the current release about how the odd mages in the mountains or whatever "knew better" than to bother them? There previously were scenarios at that point. Available units change, too; you used to be able to recruit gryphon riders after crossing the river, and now you can't recruit them until you go into the caves. (This *greatly* changes the dynamic in the mountainous "we must fight for a chance to rest" scenario.) And there's a pending change (well, pending last I checked) that introduces an "arcane" damage type, which replaces holy in some cases and cold in others (cold will still be a damage type, but units like the Dark Adept for instance won't use it any more), and all the unit resistances and scenario balances have to be rejiggered to take account of that...

I'm not entirely sure I fully agree with the decision to not support saved games from earlier releases, but I would be hard pressed to argue coherently that they *should* be supported. It would be difficult to support them well, and anything short of bundling every old release with the new release (and having the open-saved-game routine call the appropriate version of the game) would be sure to have significant problems.

I definitely agree, however, that the game should have at least informed the user, the first time on seeing them, that "unfortunately, saved games from earlier releases are not supported", or words to that effect. They won't like it, but at least they won't bang their head against the wall uninstalling and reinstalling and trying to get it to work somehow.

One of the worst usability gaffes I've encountered is an application featuring non-resizeable windows that are significantly too narrow to display the (tabular) content that is placed in them. This is in a line-of-business application, so it's something most of my coworkers use for the bulk of their shift on any given day. Columns can be resized, but there's no suitable size to make them that doesn't hide something you need to see, and they go back to the poorly-chosen default widths any time you do anything much (like, open up the information on a different account). Personally, I consider it to be bad API design on the part of the OS that there even *is* such a thing as a non-resizeable window. Application developers shouldn't even have that option, as far as I'm concerned. But off the top of my head I can't name a graphical operating environment that enforces this, unless you count tiling window managers like ion and ratpoision, and even there they just pad the extra space with nothingness; the content portion of the window does not expand like you would want.