|
From: Iain S. <iai...@ya...> - 2001-04-02 20:06:43
|
Sorry. I was sick and offline this weekend! Feelin better now though. At 08:34 PM 3/30/2001 -0500, Todd L. Miller wrote: > Iain, I just realized that by putting the topic box / search link >in the third bar down, you're preventing me from using that space for the >'am I logged in or not' links. Also, the width of the menu bar to the >right means we waste a lot of space for longer articles. Furthermore, >'hot spots in the wiki' doesn't make sense if one's already in the wiki, >so I changed it to a link to the Wiki WebHome. > > So. Should I just drop the browse bar entirely, since all of its >functions, except for the link to the <webname>WebHome (and since we don't >seem to be using webs very much) are duplicated elsewhere, and replace it >with the user links? > > The other thing: with the big header and footer and sidebar, do >you think that the amount of orange and yellow is getting a bit >overwhelming, especially on short pages? Maybe restructing the sidebar >(more vertical, one entry/line) would be helpful, or maybe it should be >done in a different color scheme? (Like the tables?) Yes. I've been looking around and doing some sketches (that's about all I could do this weekend since my body hurt too much to sit at a computer). I want to do the following: One simple (and much smaller) header. The header will have the logo, and a simple bar with one line of major categories, and a second line with sub-categories for them. The wiki will be a subcategory of the documentation main category. So, for the entire wiki, you will use one, static JOS header (that ties the wiki into the rest of the jos site). I'll get the header to you as soon as I have the time to make it. In the meantime, you can just assume there is a standard, JOS header at the top as a single table that you don't have to change anywhere in the wiki. I think the best for the wiki, is to add another "bar" under the standard header that contains the wiki search, "goto topic", and login links. Finally, under that bar, I would add another horizontal bar that contains say 5 main category links or something like that. Or maybe links to each main web, WebHome (listed as the webs themselves). We do have crosslinking between webs right (one WikiName in one web, will autolink when in another web)? So a page in "intro" web to JosKernel will autolink to the actual page in the "dev" web if it exists. Or is the behavior otherwise? And what happens when mulitple pages in multiple webs have the same name? If web crosslinking works, then we should be able to use the webs to help organize content and reduce the need for the navigation bar. The reason the nav bar was added in the first place was it was too hard to move around the wiki so if we can make the wiki easier to navigate, then we won't need it. So, finally, I think it may be best to simply eliminate the left navigation bar, and put that information in the main webhome for each web. I think it may be also possible to move the links at the bottom of the page (the footer links) up to the top and make a single, compact "control panel bar" as that top, wiki bar. Then the bottom only contains the copyright and the sourceforge logo. If you want, I can mock up what I have in mind and send it to you. > Anyway, wiki.jos.org is now running the latest sfWiki code out of >the CVS, which means it should be hunky-dory, except that it can be >spider-bombed. Iain, I can't find the link I seem to remember you >forwarding me about how to take care of that. Could you send it again? Sorry. I can't seem to locate it. I guess you'll just have to start looking at sourceforge.net/projects/twiki and browse around. It was actually the cron job that ward cunningham uses at the original c2 wiki to prevent this problem (if that helps the search). It may be OK to simple put up a decent robots.txt file and monitor what goes on. Since your wiki is database backed (rather than the old file-based wiki) and searches aren't huge cpu drains like it was on the old jos wiki we may not have the same problems. I imagine that if a spider starts wacking on the sourceforge machines, sourceforge itself may have automatic defenses (preventing a single address to request pages at too fast a rate). To stop my babbling, if you can find the spider bomb prevention script, then great. Otherwise, I think we can go with the system, and develop our own if the need arises. If I recall, his cron job was simply every X minutes, check the web logs for too many requests from the same address, if there are, then block that address from requests for x hours. I'm not a unix guru but if you are or can find one, I bet they could write that script in a minute or two. -iain |