#4 bug in TA_HistoryAlloc
TA_HistoryAlloc returns internal error 5044 when
TA_HistoryAllocParam.end is set to a date more than 2
days less than the ascii file end date.
Example:
My yahoo downloaded ascii file has the following
records:
ERS.csv
Date,Open,High,Low,Close,Volume,Adj. Close*
10-Nov-05,59.92,60.60,58.93,60.46,3296000,60.46
9-Nov-05,59.29,59.99,58.87,59.70,3117300,59.70
8-Nov-05,59.63,59.96,59.16,59.38,2307100,59.38
7-Nov-05,59.76,60.13,58.99,59.93,2319700,59.93
4-Nov-05,59.92,60.68,59.37,59.93,3616800,59.93
3-Nov-05,60.65,61.07,59.90,59.95,3550800,59.95
.
.
.
The following works fine:
histparam.symbol="ERTS";
histparam.period=TA_DAILY;
histparam.start=20041101;
histparam.end=20051110;
retCode = TA_HistoryAlloc( db_, &histparam, &data);
However, the following exits with error code 5044.
histparam.symbol="ERTS";
histparam.period=TA_DAILY;
histparam.start=20041101;
histparam.end=20051108;
retCode = TA_HistoryAlloc( db_, &histparam, &data);
I am using 0.1.5 cvs sourc on linux..
I spent a few hours going through ta_history_builder.c
but could not track it down.
You can contact me at pph@crisp.net if more
information is needed.
Thanks,
pph
Followups:
Date: 2005-11-15 23:51
Sender: pph1
Logged In: YES
user_id=1379194
My firefox must have misbehaved with double posting.
This issue is not fixed with the latest chagnes. Let
me know if you need any more information.
Thanks,
pph
-----
Date: 2005-11-15 19:49
Sender: nobody
Logged In: NO
Thanks Mario for your quick reply.
I updated with the latest changes (which had updatesin
ta_data subsystem), and this seems to have fixed the
issue!
BTW, I am not using 64 bit platform.
On a side note, is there any document on enabling
tracing
and debugging while using the debug version of the
library?
This would save me some aggrevation and might be able
to
provide more data (or a fix) if there is another
problem I
encounter.
Thanks again for sharing the wonderful library.
pph
-----
Date: 2005-11-15 18:44
Sender: nobody
Logged In: NO
Mario,
Sorry, spoke too soon. I had made a workaround to this
problem and while testing it last time, it was still
in effect.
The changes did not fix the issue. Got same 5044
error when the end date is <= 2 days before the last
day in the data source.
Thanks,
pph
-----
Date: 2005-11-15 18:34
Sender: nobody
Logged In: NO
Mario,
Forgot to answer your second question. I am using
yahoo daily data in one datasource and incremental
updates in another data source. However, the same
problem persisted when I used only the first data
source.
pph
-----
Date: 2005-11-15 18:29
Sender: nobody
Logged In: NO
Thanks Mario for your quick reply.
I updated with the latest changes (which had updatesin
ta_data subsystem), and this seems to have fixed the
issue!
BTW, I am not using 64 bit platform.
On a side note, is there any document on enabling
tracing and debugging while using the debug version
of the library? This would save me some aggrevation
and might be able to provide more data (or a fix) if
there is another problem I encounter.
Thanks again for sharing the wonderful library.
pph
-----
Date: 2005-11-14 03:31
Sender: fortier
Logged In: YES
user_id=1123
Thanks pph for reporting this.
I have a few more questions:
(1) Is this a 64 bits platform? If yes, we have
another contributor that did identify issues related
to timestamps on 64 bits platform only. We are
working on it.
(2) This internal error is in the merge logic. Is it
intended that you have two datasources for the same
category/symbol? If yes, please provide me info on
that second data source.
Pawel did check-in changes around the same day you
did report that bug. Perform an update of your local
copy to make sure you got all his changes.
Thanks,
\Mario
