I have a graph here with temperature on the X-axis and pressure on the Y-axis for several part numbers. I'm running into the above error...


The graph is defined as:

GRAPH FILE DATA
-* Initially Created by Advanced Graph Assistant
PRINT 'E''' AS 'E'' (MPa)'
BY 'PART_NO' AS 'Part no.'
ACROSS 'T' AS 'T (°C)'

ON GRAPH SET HAXIS 19
ON GRAPH SET VAXIS 12
ON GRAPH SET UNITS CM
ON GRAPH SET LOOKGRAPH SCATTERS
ON GRAPH SET GRMERGE ADVANCED
ON GRAPH SET GRMULTIGRAPH 0
ON GRAPH SET GRLEGEND 1
ON GRAPH SET GRXAXIS 1
ON GRAPH SET GRAPHSTYLE *
setLogScale(getY1Axis(),true);
setConnectScatterMarkers(true);
setScaleMinAuto(getX1Axis(),false);
setScaleMin(getX1Axis(),-100.0);
setScaleMaxAuto(getX1Axis(),false);
setScaleMax(getX1Axis(),100.0);
setX1MajorGridStep(20.0);
ENDSTYLE
END

Is there a way to tune that limit on ACROSS up (I'm not going over 255 by much) or is there a way to turn that ACROSS into a BY and keep the grouping by part no. for the legend?
The Y-axis is logarithmic, I don't think I can easily swap the series around to the other axis?

This message has been edited. Last edited by: Wep5622,

? 32451

0(FOC32451) Max of 255 across instances in online report exceeded
This limit does not apply to HOLD files. Consider generating a
HOLD file instead.

I can't, this is for a GRAPH request.

PRINT --> Y-axis
ACROSS --> X-axis
BY --> Legend

I tried with two BY-fields, but either way around doesn't result in the desired graph.

I know of one workaround, which is to:

• write a hold-file with the Y-values per legend in a column each, which creates a variable number of columns,
• then using CHECK FILE HOLD to inspect the resulting table definition,
• writing the relevant column names to a text file, and finally
• include that text file in the GRAPH request, but ye' gods!

There has to be a simpler approach!

This message has been edited. Last edited by: Wep5622,

Perhaps I didn't put enough emphasis on the fact that this graph has an X-axis, as opposed to an Ordinal axis.

There is a distinct difference between those two, namely: "An ordinal axis is a type of axis in which the actual values of the points are not significant and not used."

I am NOT using an ordinal axis.
The actual values of the points on our X-axis matter and they have enough precision that we can have any number of different values on that axis, although they are all in a range of -100 to 100.

A limit of 255 values is preposterous in our case.

Nobody knows the answer?

There's always the old method of extracting the data into save files and then including them within the graphstyle syntax such as within this post by JG.

T

My recollection is that your could change ACROSS to BY, and still force the desired graph type by means of explicit ON GRAPH SET specifications. Try it.

Yes, and I have used that in the past.
However, in this case it changes the graph! Instead of multiple curves per PART_NO I get a single line that follows X==Y.
The only reason the remaining line is curved seems to be the logarithmic scale of my Y-axis.

Too bad we can't upload images to this forum or I'd post the results of both versions.

Save your image an a public website (e.g. webshots or minus) and post the link.

I opened a case with IBI instead, as the images are not supposed to be publicly available.

To demonstrate what Wep infers, use this code.

Wep, if this code doesn't show the difference that you mean please let me know and I'll delete it

DEFINE FILE GGSALES
  XAXIS/D5 = INT(DOLLARS / 90) - 150;
  YAXIS/D5 = INT(BUDDOLLARS / 100);
END
GRAPH FILE GGSALES
 PRINT YAXIS
    BY PRODUCT
ACROSS XAXIS
WHERE PRODUCT IN ('Capuccino','Espresso','Latte')
-*ON GRAPH SET GRAPHEDIT OFF
ON GRAPH SET HAXIS 950
ON GRAPH SET VAXIS 600
ON GRAPH SET UNITS CM
ON GRAPH SET LOOKGRAPH SCATTERS
ON GRAPH SET GRMERGE ADVANCED
ON GRAPH SET GRMULTIGRAPH 0
ON GRAPH SET GRLEGEND 1
ON GRAPH SET GRXAXIS 1
ON GRAPH SET GRAPHSTYLE *
setTextString(getTitle(),"X Axis as Across");
setDisplay(getTitle(),true);
setLogScale(getY1Axis(),true);
setConnectScatterMarkers(true);
setScaleMinAuto(getX1Axis(),false);
setScaleMin(getX1Axis(),-100.0);
setScaleMaxAuto(getX1Axis(),false);
setScaleMax(getX1Axis(),100.0);
setX1MajorGridStep(20.0);
ENDSTYLE
END

I'll publish the images as well soon.

T

Images -




T

Yes Tony, I think that's pretty similar to what I'm seeing. Everything ends up at the positive side of the X-axis, at least, same as in my case.

I also seemed to see a strong correlation between values on both axis; if not displayed on a logarithmic Y-scale it was almost a straight line! There were some odd discrepancies in that line though: sometimes the lines between 3 subsequent points formed a 90 degree angle. I'm not sure how much of what I witnessed was due to our particular data.

I'm curious, if you put that on a linear Y-scale, what would it look like?
I'd try myself, but we removed (among others) the GGSALES sample file from our installation.
There's probably some way to "re-install" it. I recall we did that for the CAR database, but I can't recall how and searching the forums doesn't yield the answer...

Like this -



PM coming your way regarding the GGSALES DB

T