Detailed error messages

[cap]

Hello,

I would like to request some manner of providing the user with error messages. For example, there have been numerous times where I have made pure syntax errors, but just didn't know, and instead thought something was wrong with my data, or other times where I was just left with no explanation at all (eg, trying to make scalar variables in the command window). The status bar messages on the bottom of the screen are a nice start, but that hardly allows for decent space to give the user a fair amount of information, or the ability to review previous error messages.


Thanks in advance,
-cap

[EViews Gareth]

Its a general EViews philosophy that invalid options are ignored, rather than causing an error message. The advantage of this is that you can, for example, switch estimation types very easily without having to worry about whether an option is still valid for the new estimation type (since if it isn't valid it will just be ignored). The downside is, as you point out, if you make a mistake you might not learn about it.

The particular example you give of trying to use a program variable outside of a program (I presume you mean a !i type variable by "scalar variable") is one where we could error but currently don't. I'll investigate adding the error message.

With regards to error message history reporting; we're working on something along these lines.

[cap]

Ok, I see what you're saying, but I don't think our two ideas are mutually exclusive. For example, upon reading some syntax mistake or having an invalid option specified, instead of generating an error, a non-fatal warning could instead be written. In particular, I imagine some sort of text log object that could be standard to all workfiles (like c,beta, and resid), and the text in it could be updated after every user action.