Long time no code, but soon there will be

Talk about random things WiFiDB related, either new things, concepts, or issues to work out in code.
Post Reply
User avatar
pferland
Contributor
Contributor
Posts: 406
Joined: Mon Oct 22, 2007 8:38 am
Location: The Universe
Contact:

Long time no code, but soon there will be

Post by pferland » Tue Jul 07, 2009 11:34 am

Ok, so its been a real long time since I made a release. Between trying to square away the annoying GPS signal pairing issues and other small annoying bugs, a stolen laptop that put me back two weeks, and other bugs return after something else is fixed, or ill just screw a whole set of code up. I am hoping for some time this week to be finished with 0.16 Build 3. Some of the fixes so far:
  • 1. Fixed an issue where if you had sent a file from vistumber the token would not be able to be compared.
    2. Added Token support on almost every page that takes user input, or input from the URL.
    3. Added signal strength to Access Point Signal Plot KML export.
    4. Changed Individual User Stat Page to new layout.
    5. Standardized SSID usage, there are three types:
    • 1. <ny-paq>|sayz oh-my this is fun ....... [ is the un-sanitized, but safe ssid, used in the pointers table, 32 char limit, no special chars.]
      2. <ny-paq>_sayz oh-my this is fun ........[safe for file names, no special chars, 32 char limit.]
      3. _ny-paq__sayz oh-my this ...............[safe for table names, max 25 char, no special chars.]
    6. Changed the way WiFiDB looks for the install folder.
    7. Daemon now has all SQL based errors being echoed out to the screen.
    8. Changed the table hide for GPS history, so it is now hidden by default. Has a +/- symbol to either expand or contract the table next to the GPS History Title.
    9. For some reason the Save Search link was missing after 0.16 Build 1, is now back, with a few enhancements.
    10. Fixed the No Token error with imports directly from Vistumbler.
    11. Fixed the formatting of the file location that is in the URL from the import directly from Vistumbler that gets printed on the page.
    12. Also made the drop down for selecting the refresh time on the scheduled imports page, so it has a default of the current selection instead fo going back to 5 sec.
    13. Fixed the Upgrade script in /install/upgrade/patch.php. (Was broken in Build 2 and 2.1.)
    14. Fixed the Install script in /install/install.php. (Was broken in Build 2 and 2.1.)
    15. Added Daemon Status to the scheduling.php page.
    16. Fixed an issue where the Signal history was being corrupted by being sanitized. The 'special' characters '-' and ',' were being encoded.
    17. Fixed Some Issues with the daemon, details are in the Tools Readme.
    18. The WiFiDB Web log vars $log_level and $log_interval where interfering with the Daemon vars, they are now changed to $log_level_W and $log_interval_W.
    19. The Username in the 'Files already imported' table was pointing to the All APs for that user page, when it is supposed to point to the Users stat page.
    20. Fixed an issue where if some AP Pointers are removed from the `wifi0` table, some APs will fail to import or have the same ID as another AP and not be linkable to their data.
    21.
    • [ Issues reported by ACalcutt ]
      1. "Access" is spelled incorrectly multiple times on Export page.
      2. (Export an Access Point to KML) "Username" should be "SSID".
      3. Fix the links for Access Points on the All AP page. (Add tokens to the SSID URLs in the "View all APs" page.)
      4. Refresh time on the scheduling page went to 15 seconds when I set it to 5 seconds. On the next refresh it went back to 30 seconds.
    22. Unified the import_vs1() and importvs1d() functions, added an $out var to import_vs1() and verbose(), valid values are "CLI" and "HTML".
    23. Moved the Install folder warning code to the database.inc.php file from config.inc.php.
    24. All Messages in import_vs1 are in a group of variables in the beginning of the function, for easy editing.
    25. Added in some code to handle obscure APs that get tagged as new when they are not new.
    26. There was missing Token links on the Associated Lists section of the AP Fetch page.
    27. Numerous other small fixes that I have forgotten about.
The Production server (http://www.randomintervals.com/wifidb/) has been in rebuild mode (Not accepting imports) for some time, due to the fact that it has the old broken importing script on it and was continuing to sceew the GPS/signal pairing. although i did end up creating two new tools called:

filename_create.php - To grab all the files in /import/up/ and compare them to the `users` table and create a file called `filenames.txt` with all the relevant data too be imported into a new DB for rebuilding

daemon_prep.php - Takes the filenames.txt file and inserts it into the `file_tmp` table for the daemon to import.
The best acceleration you can get on a Mac is 9.8ms^2

User avatar
pferland
Contributor
Contributor
Posts: 406
Joined: Mon Oct 22, 2007 8:38 am
Location: The Universe
Contact:

Re: Long time no code, but soon there will be

Post by pferland » Tue Jul 07, 2009 3:08 pm

ok so the import has been fully fixed, there are no erroneous GPS ids that have no meaning or 0,0's
also if for somereason the import script cannot find the ap in the pointers table, and it thinks its a new ap, but its really not (becasue teh signal and GPS tabels exist) it will use the creation of the signal and GPS tables as a flag, if both the signal and GPS tables fail to be created, it takes this as an assumption that the AP already exists, and continues on as if it were an update to an AP, gos and finds the APs real UID and assigns it that ID.

say we have this AP:
SSID: linksys
MAC: 00:22:6B:53:6D:81
SECTYPE: 3
RADIO: g
CHAN: 6

and when i select from the pointers table, it would return no rows so this means it thinks this AP is new.

so we go and try to create two tables `linksys-00226B536D81-3-g-6` and `linksys-00226B536D81-3-g-6_GPS`

if `linksys-00226B536D81-3-g-6` fails and `linksys-00226B536D81-3-g-6_GPS` fails this is an assumption that the tables are already there, because we could also get valid returns from MySQL not connection failures to get this far.

if `linksys-00226B536D81-3-g-6`was successful but `linksys-00226B536D81-3-g-6_GPS` was not, this is a problem and the script halts with the error.


For any that are interested, I have a full import going now at http://rihq.randomintervals.com/wifidb/ ... duling.php
as of this post it is on file 24 of 282
This is the QA/DEV server so after the testing is done, its being moved over to PROD and rebuilt with valid good data

i think this is a good example because EIHOME has a lot of signal history and in the end it will have somewhere around 31 history rows.
The best acceleration you can get on a Mac is 9.8ms^2

User avatar
pferland
Contributor
Contributor
Posts: 406
Joined: Mon Oct 22, 2007 8:38 am
Location: The Universe
Contact:

Re: Long time no code, but soon there will be

Post by pferland » Sat Jul 11, 2009 11:06 am

OK so i have all the code for 0.16 Build 3 finished.
The Release post is over here:
WiFiDB Alpha 0.16 Build 3 Release
The best acceleration you can get on a Mac is 9.8ms^2

Post Reply