skuzzle / polly Goto Github PK

Branch 'new_ast' should be merged into master soon

https://github.com/skuzzle/polly/blob/master/polly.sdk/src/de/skuzzle/polly/sdk/time/DateUtils.java#L230

does not work correctly

we currently do not distinguish between integers and floats, but some operators (like shifting) require integers. Currently, the float number is just rounded to get the integer, which may be inaccurate

...is even messier as before the "improvments". Things issues/missing/wanted:

• compile and export a single plugin. Especially in a way that polly can then be debugged using the built plugin
  (* maybe even better debugging support which does not need to run the build file for a plugin or eclipse should run the build file)
• compile and export only plugins
• consider replacing auto eclipse project creation with fixed projects (due to missing features in Ant Eclipse Task)
• general inteference of ant build with eclipse build need to be fixed. Maybe eclipse should run Ant files instead of internal builder

Captchas marked by polly for manual classification should be classifyable using the webinterface

Positions of number literals that start with a '.' (like .1234) are not correct, because an artificial '0' is pushed back during scanning.

Create an overview of alien races, spawns, average drop and damage

Hashcode of the captcha image might be used to determine whether the current code is still valid

Plugins should be able to report recent messages which will be shown to all users within the webinterface

The current buildscript is very shitty and contains massive duplicated "code". That should be changed by intorudcing reusable ant targets.

Fix that damn encoding. Convert ALL files to UTF-8.

IP session identification is pretty crappy und should be replaced with a system using cookies

Until now, all 15 surrounding scoreboard entries are persisted within one transaction => that's shit!

It does not produce the same result as the HashMapTypeUnifier

need to be set

iframe/ajax whatever

Deadlock somewhre in these classloaders leads to polly being totally blocked

The persistence.xml is generated during program start by polly.When running from eclipse, it is generated within the wrong folder but is copied by eclipse to the bin folder. Thereby the persistence.xml within the bin folder is always the version generated by the last run of polly and not by the current.

[15:49:20] @clum wtf
[15:49:24] @clum bei der umgebungsliste
[15:49:40] @clum habe gerade nach einem venad gesucht
[15:49:43] @clum venad angeguckt
[15:49:46] @clum filter gelöscht
[15:49:50] @clum anderen venad angeguckt
[15:49:58] @clum und der plot ist auch für den neuen venad
[15:50:06] @clum aber die angezeigten einträge sind vom vorherigen

E.g. for buildings

[21:30:20] <!qzone> betty: http://qzone.servebbs.net/pollyPriceDaily.php?date=2014-02-10
[21:31:28] @C0mb4t Was passiert wenn es für das Datum keinen Eintrag gibt?
[21:32:20] <!qzone> id;1 date; nrg; rek; erz; org; synt; fe; lm; sm; em; rad; es; eg; iso;
[21:33:31] <!qzone> @C0mb4t Was passiert wenn es für das Datum keinen Eintrag gibt? <- ab jetzt: id;0 date; nrg; rek; erz; org; synt; fe; lm; sm; em; rad; es; eg; iso;
[21:34:04] <!qzone> also falls 0 einträge: id;0

Rx in its peculiar ways places ship attributes with a null value different from non-zero attributes in battle reports.

Frachtschiff (LXXVIII) Getza    Recruit Fabiana Syberg
Angriffswert    0/0     Captain -559 / 6(-565)  XPs 0
Schild  0 / 260(-260)   Crew    9 / 62(-53) XPs 0
Panzerung   0 / 89(-89)     Systeme 28 / 417(-389)
Struktur    0 / 4(-4)   ID:16223490 
zerstört

Note the whitespace for "Angriffswert". Although it doesn't show in a copy & paste, there actually is another space character in front of 0/0. The spacing makes at least the regex in https://github.com/skuzzle/polly/blob/master/projects/de.skuzzle.polly.plugin.rx/src/polly/rx/parsing/BattleReportParser.java barf...

The new web interface should allow users to define remind templates which can be easily scheduled using either irc or the web interface

public String toString(boolean printPosition, boolean printValue) {
    StringBuilder result = new StringBuilder(15);
    String value = this.stringValue;
    switch (this.type) {
        default:    
        case DATETIME:
            value = this.dateValue.toString();
            break;
    }

...

PersistenceManagerV2 currently uses one EntityManager instance during its whole life time. As this persistence context grows it slows down the whole application. This model must be changed to create a new EntityManager for each transaction and read access.
This will cause all entity objects to become 'detached' after the database operations which will possibly break a lot of existing code

While fixing encoding like in ticket #24, all strings should be externalized using proper technics like ResourceBundles for example

only reminds that the user created, otherwise it will delete a remind created by another user.

As we have a webserver now, polly might as well use it as a paste service

One of the oldest modules which should be rewritten to provide some more features (for Example storing command aliases persistent)

parsing of "list<list>" fails, because ">>" is recognized as a right shift

Considering that we lost the last Robocup in Mexico, it would be helpful to implement more cool stuff like obstacle detection etc...

Webserver does currently know nothing about content types: request body is always interpreted as String AND as formular data. Encoding specified by the requesting client is also ignored
There is much room for improvements

Plugins should be able to report all kinds of status messages which will then be shown on the corresponding admin web page

On:
[11:32] me: :cracker
[11:32] polly: Danke für den Cracker! Das war der 5. Cracker für me

the correct answer would be

[11:32] me: :cracker
[11:32] polly: Danke für den Cracker! Das war der 5. Cracker von me

HTTP server currently uses hardcoded ISO encoding whenever receiving or sending data. As polly always uses UTF-8 this causes display errors on the webpages

Currently, expression types are only resolved when a variable was stored but not again when a variable is used. This must be changed as the usage of a variable may impose a different type context than its declaration.

Server side cell editors know about different content types of cells. The client side java script does currently not know about those types and interprets everything as a string which is unflexible and doesn't even work for most types.

Class should be moved (and extended by functionality) after new HTTP branch has been merged into master

as it says, we need the remind-ID in the snooze-message

[16:19:29] @nighthawk im Scoreboard 5 Venads verglichen, dann die Darstellung auf 8 Monate verk�¼rzt
[16:19:49] @nighthawk anschliessend 5 andere Venads gew�¤hlt, und wieder die Darstellung auf 8 Monate gek�¼rzt
[16:20:00] @nighthawk dabei wurde dann das alte Diagramm gezeigt
[16:21:14] @nighthawk hab dann mit dem Browser nen Refresh (F5) gemacht, und nochmal von 12 auf 8 Monate ge�¤ndert
[16:21:18] @nighthawk dann klappte es