Testing 'CopyTicks' - page 34

[[Deleted]]

coderex:

Doesn't the exchange provide data on who exactly did the transaction? Is there an ID of some specific counterparty

[[Deleted]]

coderex:
well why nothing interesting, there just for a fee the information is valuable - a tape of trades, for those who test stochastics on history such information is very valuable, it is a pity that in MT5 the tester does not take this history

I judge by what I see in the test data:

#SYMBOL,SYSTEM,TYPE,MOMENT,ID,ACTION,PRICE,VOLUME,ID_DEAL,PRICE_DEAL
RIM5,F,S,20140205184842717,12090675286,1,151990.00000,1,,
GZH4,F,B,20140205184842717,12090675287,1,12765.00000,1,,
GZH4,F,B,20140205184842717,12090675288,1,12717.00000,1,,

[[Deleted]]

Alexey Kozitsyn:

I judge from what I see in the test data:

#SYMBOL,SYSTEM,TYPE,MOMENT,ID,ACTION,PRICE,VOLUME,ID_DEAL,PRICE_DEAL
RIM5,F,S,20140205184842717,12090675286,1,151990.00000,1,,
GZH4,F,B,20140205184842717,12090675287,1,12765.00000,1,,
GZH4,F,B,20140205184842717,12090675288,1,12717.00000,1,,

Ah, stop, or is what I've highlighted just a change in the glass?

[fxsaber]

coderex:
Isn't she for yesterday?

No.

[coderex]

fxsaber:
No.

Really not yesterday's, well that's the latest data unfortunately. We will have to see on Tuesday what the latest data will be.

[[Deleted]]

This is a funny picture on BCS #1 build 1470. The candlestick control takes place after the candlestick closes. The volume by ticks is compared to the volume Volume[] of the candle from the indicator. As you can see the control has changed AFTER the candle closes!

2016.11.14 14:52:02.507 VolumeControl Si-12.16: ОШИБКА 2016.11.14 12:51! Сумма V buy = 2728, сумма V sell = 4187, контроль (покупки+продажи) = 6953
2016.11.14 14:52:15.365 VolumeControl Si-12.16: ОШИБКА 2016.11.14 12:51! Сумма V buy = 2728, сумма V sell = 4187, контроль (покупки+продажи) = 6997
2016.11.14 14:52:40.259 VolumeControl Si-12.16: ОШИБКА 2016.11.14 12:51! Сумма V buy = 2728, сумма V sell = 4187, контроль (покупки+продажи) = 6997

[[Deleted]]

Dear developers(@Slawa)! Is the work on CopyTicks() bug fix going on?

[fxsaber]

Not sure how to reproduce. Appears during ZoomPrice stress work - many charts with indicator, switching symbol on each chart, changing account without unloading indicator. I.e. there is intensive work with CopyTicks.

It manifests itself like this.

Market Watch freezes for a few seconds, at this point CopyTicks does not return new data. Then Market Watch comes to life - shows fresh data. With this CopyTicks gets all of the history, which was during the pause. Then Market Watch freezes and goes round and round.

It is interesting, that when Market Watch is paused, it does not show current time immediately, but quickly (visually noticeable) scrolls all missed ticks.

Captured on video

The yellow ones are the fresh tick history. You can clearly see the Market Watch slowing down. And when it comes out of the brake, the tick history retroactively changes. That is, when Market Watch slows down and through the ticking history you cannot get data.

[coderex]

fxsaber:

Not sure how to reproduce. Appears during ZoomPrice stress work - many charts with indicator, switching symbol on each chart, changing account without unloading indicator. I.e. there is intensive work with CopyTicks.

It manifests itself like this.

Market Watch freezes for a few seconds, at this point CopyTicks does not return new data. Then Market Watch comes to life - shows fresh data. With this CopyTicks gets all of the history, which was during the pause. Then Market Watch freezes and goes round and round.

It is interesting, that when Market Watch is paused, it does not show current time immediately, but quickly (by eye noticeable) scrolls all missed ticks.

Captured on video

The yellow ones are the fresh tick history. You can clearly see the Market Watch slowing down. And when it comes out of the brake, the tick history retroactively changes. I.e. when Market Watch slows down and you can't get data through the tick history.

it seems that you have the same problem - slows down when rendering graphics.

This is where splitting the user program into threads can help:

- execution thread of the main program algorithm

- a rendering graphics thread

The possibility to create separate windows for user-specified drawing without loading history data into them, but in a separate thread of the total execution of the user program would also help.

[fxsaber]

coderex:

It seems to be facing the same problem - slowdowns when rendering graphics.

Not really. I take the indicator off, the brakes continue for a while.