Profile

outlier_lynn: (Default)
outlier_lynn

January 2015

S M T W T F S
    123
45678910
11121314151617
181920 21222324
25262728293031

Most Popular Tags

Expand Cut Tags

No cut tags
Monday, September 20th, 2004 10:29 pm
This became long and it's entirely geeky, so it's behind the cut to protect the easily bored. :)

I wrote some code today that I'm hoping I'll never had to tinker with again. It works perfectly, but by tomorrow, I'm going to have no clue how it works. :) It's doing a list search on list a, using the idx to find a file name from an array, then using the file name to find yet another index from another array, then checking that the index can in incr or decr without failure, then providing the id of the file for yet another index to return a directory and file name for correctly displaying a jpg. whew. :)

I'm have some fun now creating views and such in postrgresql. My gallery scripts are nearly complete. The tough parts are almost done. I'm providing an easy way to get an image from the gallery into the front page rotation. Well, easy for the user, anyway. :) I'm having to backfit the front page scripts. It suddenly came to me that it would be too easy to put the same picture on the front page rotation more than one time. Not good. It will always be possible to do, but I don't want to have the gallery editing scripts do it.

I can now display an index of galleries and a special page for editing a given gallery. The only thing left to do for that part is write the wrapper around updating the database and copying the designated files to the front page.

That leaves two chunks left to complete. 1) adding a new gallery to the database. 2) adding pictures to the gallery. I think I'll write some sanity code, too. Something to be run by cron in the middle of the night. Just to check that the database and the gallery directories are in sync. That means I'll need some way to compare and find the offense, but that isn't too hard.

I discovered one of my tools didn't work if I used a frame with a default black background. All the "data" in my test page was black on black. I've been meaning to rewrite that tool anyway. It was a fast hack two years ago. And it had problems. Now it's all in native tcl. The only packages it uses are mine. ;)

Later this week, after I have the CI scripts done, I'm going back to work on the mysql tcl package. It's weak and NGender should have a package ready.

This week and probably next will be busy. A second hardware box will be set up and all the server stuff will be migrated to it. That will be a bit harder than it might seem because the server is several generations old in both mysql and postgresql. In the development work I've done in the last two weeks, I've had to tweak the sql dumps on Peace (my desktop) to get them to create the relations I needed on hermes (the server). And I want to migrate a bunch of stuff on mysql off it to postgresql. That will be fun. I'll play with it on Peace until I feel confident in an automated solution.

I'm getting much, much faster at finding decent solutions to tight spots in database and code design problems. Yay me. :) I'm still have to look up every other command I'm using to get the syntax right though. I've started have [livejournal.com profile] _greg's disease of forgetting which language I'm writing in. :) Funny enough, I'm making mistakes in tcl because I start writing C. I haven't written anything in C in years and years.

I have gotten very tired of having mistakes on the CI menus from one page to another. I was very slow at doing anything about it, though, because it's apples and oranges and bananas. There is tcl, pl, php dynamic pages and fixed pages. And the menus aren't quite the same on some of them.

Late last week, I gave up and started learning php. Now the tcl and php dynamic pages get their menu from the same file. Whee. The same mistake will be everywhere now. :)

I thought about fixing the perl scripts to grab the same file -- that's an easy fix -- but I'm going to dump the perl anyway. I've written all the functionality of the perl scripts into tcl using the postgresql database. When I'm through with the gallery editing scripts, I'll whip up a script to install all the galleries from the old system. I'll leave the images in the old locations for a bit, though, so links people have used won't break.

There is a strange, strange problem at E Street Cafe. Greg and I are going to look at it tomorrow, probably. The firewall system has a red, green and blue component. (The blue isn't really all that secure since it's the public access net in the cafe.)

However Mordred is listening to six or seven real world IP addresses and is plugged into the green net or DMZ. It's listening to all those IPs with the same NIC. So far so good. One of the addresses serves the E Street Cafe web pages. And it is possible for all the world to point their browser to estreetcafe.com and see the pages.

All the world except for anyone on the blue net or anyone else on the green net. I'm especially interested in why my server, plugged into the green hub, can talk to Mordred, also on the green hub, on most, but not all, of it's IPs while requests coming from the RED world have no problem.

I have two guesses at the moment. One is a swag and really can't account for this behavior. I'll check that one out as a last resort. :) The other guess is that Mordred (a Microsoft server) has some firewalling going on and is rejecting any IPs from it's local network. The blue net is natted so it will look green to Mordred. Hermes is green. And the rest of the world will reach Mordred from a different net.

It accounts for the symptoms. But those who know more about the Windows2000 server are suggesting that is isn't possible.

Isn't that special.

Anyway, it will be a fun problem to tackle.