computer chip



home
archive

suggestions
help page
future features



View current page
...more recent posts

Alex pointed out that the new framed archive view was not working in his browsers (nav 4.x on Mac and I think an older IE on windows.) I believe I have fixed this issue (confirmed for 4.x on mac, but not for whatever the older IE on windows is.) Can anyone confirm this, or spot other problems? Thanks.
- jim 11-24-2001 4:18 pm [link] [add a comment]

Thanks to Alex for finding a somewhat subtle bug. If you tried to move a page by changing the path in [editpage] (which is how you would move a page) and you changed the name to a longer but otherwise similar name (from /test/test2 to, say, /test/test234567) the system would get confused. And that's putting it mildly. Should be fixed now.
- jim 11-23-2001 10:30 pm [link] [add a comment]

In [editpage] you can now switch from the 'standard template' to 'use your own HTML'. If you choose to do this you must make the change to 'use your own...' and then come back to [editpage] a second time and you will see some different options including two new posting boxes (textarea boxes) one labeled 'Opening HTML' and the other 'Closing HTML'. When your page is requested the system will print out 'Opening HTML' followed by your currently active posts from the database, followed by 'Closing HTML'.

As of now the posts from the database are printed as rows in a table (where each row has one column) so it expects a <table> to already be set, plus that it will be closed (</table>) in the 'Closing HTML' (along with </body> and </html>.) I think I'll take out the table in the future and just seperate the posts with <p> tags.

Also there is no way yet to create a 'use your own HTML' page from /create. I'll add that tomorrow. Right now you'd have to create a regular page and then change it in [editpage] (and then go back to editpage and add in the HTML!)
- jim 11-23-2001 9:34 pm [link] [4 comments]

Occasionally, for some as yet unknown reason, the system will screw up and think you have an unread post or comment on a page where you do not in fact have any unread posts or comments. I'm working on trying to make this never happen, but in the meantime I've changed the subscription script so that if you change your subscription to turn tracking off it will clear out all unreads for you associated with that page. Then you can turn tracking back on and it will start keeping track for you again. Note that this works from the individual subscription pages (so from /schwarz/subscriptions in this case, not from just /subscriptions.)
- jim 11-23-2001 6:34 pm [link] [add a comment]

Not sure how closely anyone is following along. I wrote some stuff about the discussion system here on my page recently. The gist of it is that threading (being able to reply to replies) is very costly in terms of storage. I'm commited to providing threading, and at this point the cost doesn't seem to be impacting performance (although in the future this might be a different story.) But I'd like to encourage a more linear comment style where this is possible.

Anyway, one idea I had was to put the posting box at the bottom of the comment page itself. This box would be equivalent to clicking 'add a comment' at the top of the comment page, or in other words, a post made from this box would add the comment to the bottom of the page, all the way on the left (or in still other words, it would add the comment to the bottom of the page as a top level comment.) You could still click [add a comment] on any particular comment to make a threaded comment that will go directly beneath the comment you are responding to (and be indented slightly.) But I think the convenience of having the posting box already visible on the page will make people more likely just to add a top level (non threaded) comment rather than clicking through to a seperate posting page.

Any thoughts? If not I'm going to try this out soon. We can always go back.

Also, I wonder if this would make unknown anonymous surfers more likely to comment? The box would be right there after all. Would that be good or bad?
- jim 11-20-2001 5:07 pm [link] [add a comment]

All searches (regular and advanced) will now return a maximum of 50 results. If there are more it displays the 50 and advises that there may be more and you should try to search for something more specific. This is to guard against people searching for nothing (or 'e') and taking too many system resources to return every post from a specific page (or the whole site!)

Is 50 a big enough number?

Also all searches were updated to not return 'preview' or 'pending' posts even if they are matches (already they were smart enough not to return matches from private pages - unless you are subscribed to a particular private page in question.)
- jim 11-19-2001 8:57 pm [link] [add a comment]

A few momentary outages today as I kept breaking things. When posting you can now set the post to 'pending' in addition to 'preview'. If you choose pending you must enter a publication date in the 'pending date:' box. This date has to be in the exact format

yyyy/mm/dd 24:00

Thats the full 4 digit year followed by the two digit month (01 = january) followed by the two digit day (01 for the first of the month.) Then one space, then the 24 hour east coast time (16:30 = 4:30 pm eastern time.)

The post will be invisible until that time, at which point the following page load will publish the pending post. The 'posted by' footer will contain the date and time you entered into 'pending date:'.

- jim 11-18-2001 10:32 pm [link] [5 comments]

If a page is set to notify userland (weblogs.com) you will now get a check box on the [edit] screen allowing you to selectively notify userland on [edit]s. This would be necessary, at least, if you make a preview post (which won't nofity userland, even if you have your page set to do so) and then change it to a regular post in [edit] to actually publish it. In that case you can check both the 'treat as new post' box (which will notify the front page that there is something new) and the 'notify userland' box.
- jim 11-18-2001 4:32 pm [link] [add a comment]

Adding new users to the system has never been simple. The automated scripts I wrote last year never worked perfectly, and I always had to go into the database by hand and fix things. But I didn't realize how broken it was until Steve pointed out that Julie (the last person added) had all sorts of posting powers she shouldn't (although only on pages at least two levels down in the file hierarchy.) Anyway, I dug in today to finally work all that out and I believe I was sucessful.

Many people will now see an additional option of [add user] in their menu bar. This takes you to /user where you can easily add a new user. You must supply a unique name and a valid email address. The system will make the accout, subscribe to all pages, and make a home page with the same pages a guest surfer sees (and sets the system to track new stuff on all the pages on the home page.) The new user will be able to comment but not post to all pages. Individual page owners (for /treehouse and /sustenance and /cinefiles) can grant new people posting powers on those group pages as they see fit.

The new user does not get their own page automatically.

You make the account and the system sends the information to the new account holders email address.

The fate of our little world is in your hands.
- jim 11-17-2001 10:01 pm [link] [add a comment]

I've added a fifth option to the already hard to understand /log page. Now you can view your hits by useragent (where useragent is the type of browser making the request for your page.) Probably you wouldn't keep it on this setting, but you can switch to it to uncover who is behind hits which the 'complete log' list as being from nowhere. Probably that doesn't make sense, but check it out if your curious.

For example, on my page, if I set it to complete log, I see that the overwhelming number of hits are unidentified (that is, they are not coming from a specific link on another page.) But the log is recording not just the refering page, but also the useragent (browser and OS) of the computer making the request. So if I flip the /log to 'useragent' I see that most of those unidentifid hits are coming from the googlebot. Also I can see what other browsers people are using (mostly IE.)
- jim 11-17-2001 4:38 pm [link] [2 comments]