kxsystems / cookbook Goto Github PK

The Java code here: https://code.kx.com/q/kb/timezones/ is flawed when it comes to historical offset changes.

The code uses TimeZone.getTimeZone("...").getRawOffset(), which as the JavaDoc states, only works for 'now':

This issue is seen when generating tzinfo for Europe/Istanbul - it returns +3h perpetually even though this should only be the case in more recent years (Oct 2018+); in older years it should still alternate between +2h and +3h.

Using ZonedDateTime.of(year, month, day, hour, min, sec, nano, ZoneId.of("...")).getOffset() accurately returns historic offsets. It is not clear to me why storing DST as a separate offset was required in the first place.

        System.out.println("Europe/Istanbul 2000.01.01 offset:");
        ZonedDateTime zdtOne = ZonedDateTime.of(2000,1,1,9,15,0,0, ZoneId.of("Europe/Istanbul"));
        System.out.println(zdtOne.getOffset()); // +02:00
        System.out.println("Europe/Istanbul 2000.06.01 offset:");
        ZonedDateTime zdtTwo = ZonedDateTime.of(2000,6,1,9,15,0,0, ZoneId.of("Europe/Istanbul"));
        System.out.println(zdtTwo.getOffset()); // +03:00

When reading second+ chunks from same file the the script is treating data record as header and thus skips a record.

shouldn't this portion :

// check if we have already read some data from this file
// if this is the first time we've seen it, then the first row
// contains the header information, so we want to load it accounting for that
// in both cases we want to return a table with the same column names
data:$[filename in filesread;
[flip columnames!("PSFI S";enlist",")0:rawdata;
filesread,::filename];
columnnames xcol ("PSFI S";enlist",")0:rawdata];

be:

// check if we have already read some data from this file
// if this is the first time we've seen it, then the first row
// contains the header information, so we want to load it accounting for that
// in both cases we want to return a table with the same column names
data:$[filename in filesread;
[flip columnnames!("PSFI S";enlist",")0:rawdata];
[filesread,::filename;
columnnames xcol ("PSFI S";enlist",")0:rawdata]];

Hey all,

When I start up the tick.q with sym.q and feed.q with files provided as follows:

q tick.q sym -p 5010
q feed.q

The tickerplant process prints 'length error on every update, which occurs when incorrect number of elements is passed: https://code.kx.com/wiki/Errors

I suspect that this happens when the feed process calls .u.upd

Are there any suggestions as to how to remove this problem?