Graphing transforms of variables from databases - labels

[bobshackleton]

One of our bread-and-butter tasks is to compare and graph variables between one forecast run to another, with each forecast run in its own database. To ease this process, we ensure that every variable in a database gets a displayname that includes the name of the run (e.g. CG110709b::qgdp, where qgdp is the variable name and CG110709b is the forecast run). Thus any graph comparing variables yields a nice set of labels, each specifying forecast run and variable. So far so good.

However, if we produce a transform of the variables - for example, one variable divided by another - EViews appears to dispense with the displaynames entirely, and produces a graph with labels that (1) include only the variable names (not the displayname) if the variables are actually in the active workfile/page or (2) include the entire path (not the displayname) if the variables are coming from a database. This produces uninformative labels for the variables in the active workfile and annoyingly long labels for the variables that aren't, and correspondingly reduces the size of the figure to fit the long label. (See example below.)

Could you help us figure out how to have ONLY the displaynames appear in labels of graphed series when the series are transforms, either of variables in the workfile or variables brought in from a database? This is a task we do literally dozens of times a day and it's important that we are able to produce graphs on the fly with suitably informative labels.

[EViews Gareth]

Create a frml containing the expression, set the displayname on the frml, then graph that.