Re: [MAS-discuss] A Comment and a Question

[Jim C. <jt...@di...>]

Hi Chris.

> Hi everyone,
> 
> Just started using MAS a couple weeks ago and I think
> it's excellent. However, in that time there are a
> couple things I have noticed that I would like help
> on.
> 
> First, I've noticed that there is an issue with the
> addition of new indicators and the ability to graph
> them in 1.6.6g for Windows(Other versions of 1.6.6g
> are seemingly unaffected. I must admit that I did not
> test 1.6.5, so it might exist there as well). 
> 
> If I create a new indicator for 1.6.6g under Windows,
> I am unable to graph them. Using the CLI, I can see
> that the indicator is added successfully and outputs
> the correct data, but I am unable to graph it. I have
> double-checked that the format of the .ma_client.rc is
> correct. I can add an indicator that originally came
> with MAS 1.6.6g and graph it with no problems. 

It sounds like you've taken into account the issue in the FAQ about needing
to edit the .ma_clientrc file when you create a new indicator (just in
case, the Q/A is at:
http://eiffel-mas.sourceforge.net/support/FAQ.html#charting%20GUI
under: "I tried adding a new indicator to MAS.")  So the problem is
most likely caused by something else - I suspect it is the UNIX vs. DOS
text format (specifically end-ofline sequence) issue, or another formatting
issue such as using spaces to separate fields instead of a tab.
So first I'd suggest checking the format of the lines you've added against
the syntax described in the first line of the file (
# Format: 'keyword<tab>f1<tab>f2...'), making sure 1 tab is used to
separate each field; then check the actual characters used to end the lines
you added and make sure they're the same as the original version of the
file.  (If you're not sure how to do that, just attach the file in an email
to me and I'll check for you.)  Make any needed adjustments and hopefully
this will fix the problem.  If it doesn't, please post a followup (though a
followup saying it worked would also be helpful if that's the case).

(By the way, this whole procedure points out an obvious
interface/usability issue which I hope to solve in a future release.)

[more below]

> 
> It is possible that the server is chomping the
> Indicator Name and no longer corresponds to the name
> originally given. An easy way to test this is to
> create an indicator and give it any name. Now, create
> another indicator as a One_Variable_Function and
> attempt to use the indicator you originally created as
> it's input function. I've noticed that the indicator
> name for the indicator you originally created is
> correct in the list of Indicators to add, but when you
> select it, the indicator name gets chomped. This does
> not happen to indicators that came with MAS.
> 
> As an example of the above if it is too confusing:
> 
> I am attempting to add an indicator called Testing. It
> takes input from an indicator I created called
> Bollinger Squeeze Percentage.
> (Note: Also notice that the indicator Testing seems to
> have a line feed at the end of the indicator name)
> 
> -------------------------------------------------
> Choose a name for root function that does not match
> any of the
> above names: Testing
> 's (ONE_VARIABLE_FUNCTION) input function
> 1) Simple Moving Average
> .
> .
> .
> 36) Parabolic SAR
> 37) Bollinger Band Difference
> 38) Bollinger Squeeze Percentage
> 39) No Input Function
> 38
> ? (y/n) ollinger Squeeze Percentage
> ---------------------------------------------------

OK, that looks like the classic clue that you somehow added a carriage
return or line feed character (I forget which is which - the one that
causes the output to start at the beginning of the line, but not down
one line) as part of the name.  I believe I've used a DOS window before
to create indicators and have not run into this problem (though I may
be wrong - I may have just used cygwin bash), so I'm not sure how it
would occur.  (Have any other windows users encountered this problem?
Have you found a fix?)  It sounds like you're a fairly technical user,
so you might want to experiment a little and see if you can figure out how
to assign a name without a newline in it, since that's likely the sole
cause of the problem.  If you can't get it to work in a DOS window, if
you have the energy, you might try downloading the cygwin bash (UNIX shell
ported to windows) and try it with that (just search google for cygwin).

Sorry for the funky UI.

> 
> 
> 
> Same example, except using Parabolic SAR:
> ---------------------------------------------------
> 34) Upper Bollinger Band
> 35) Lower Bollinger Band
> 36) Parabolic SAR
> 37) Bollinger Band Difference
> 38) Bollinger Squeeze Percentage
> 39) No Input Function
> 36
> Select Parabolic SAR? (y/n)
> ---------------------------------------------------
> 
> Again, this seems to be Windows only.
> 
> I also have a question about LINEAR SUM. I am having
> problems with this function. This problem exists in
> both 1.6.5 and 1.6.6g so it is likely something I'm
> doing wrong.
> 
> I'm guessing that the function LINEAR SUM is a sum of
> a value over the last n-periods. So, let's assume that
> I want to find the total sum of volume over the last
> 10 periods. I have been doing the following:
> 
> N-RECORD ONE VARIABLE FUNCTION
> (No Input Function)
> |
> |
> LINEAR SUM(Integer Value for n-value = 10)
> (No Input Function)
> |
> |
> VOLUME
> 
> With a value of 10 for the n-value N-RECORD's ONE
> VARIABLE FUNCTION
> 
> However, when I attempt to view the results for this 
> indicator of a stock, MAS core dumps.

Unfortunately, I don't have the time right now to answer all your questions
on this issue, but I can say that the core dump is very likely caused by
one of your functions or operators (likely LINEAR_SUM) being configured to
try to access data beyond the available boundaries (i.e., before the first
record or after the last record).  I know it's difficult to figure out
what's going on here, but you might try experimenting with different offsets
or whatever and see if you can get rid of the core dump.

Also, I've added a -debug option to mas that I believe is in the 1.6.6g
version (not 1.6.5) that could be very helpful to you for figuring out
what's going on.  The output will be quite verbose, I'm afriad, but if you
can figure out what it is saying, you can probably get some clues as to
what is going wrong - unless it core dumps before it gets a chance to
output anything, which would be unfortunate.

Please keep us informed of what you find.  I may have some time later to
help more on this, but for any other mas users reading this who think they
may be able to help - feel free to contribute.

A developer-oriented question: Did the core dump cause the server to
terminate?  (If so, that's something that needs to be fixed - mas should
recover and let you try again.)


Jim

> 
> I would appreciate any help anyone has on this. Is it
> something I'm doing incorrectly? Am I incorrect in
> assuming that LINEAR_SUM creates a sum of a value over
> the last n-periods? Is there another way to do what I
> want to do that I'm missing?
> 
> The only other way I can think of at the moment to do
> the the above is creating an indicator with ten
> addition operations and with corresponding
> SETTABLE_OFFSET_COMMANDS. However, I'd rather not do
> something like this as it is very inefficent. :)
> 
> Thanks for your help,
> CT
> 
> __________________________________
> Do you Yahoo!?
> Protect your identity with Yahoo! Mail AddressGuard
> http://antispam.yahoo.com/whatsnewfree
> 
> 
> -------------------------------------------------------
> This SF.net email is sponsored by: SF.net Giveback Program.
> Does SourceForge.net help you be more productive?  Does it
> help you create better code?  SHARE THE LOVE, and help us help
> YOU!  Click Here: http://sourceforge.net/donate/
> _______________________________________________
> Eiffel-mas-discuss mailing list
> Eif...@li...
> https://lists.sourceforge.net/lists/listinfo/eiffel-mas-discuss