Re: String variables combining files

[Frans Houweling]

On 26/03/2015 13:00, Alan Mead wrote:
> By "I don't think the widest variable should be saved" what exactly do
> you mean?  If I have email A32 in one file and email A31 in a second
> file, if the final merged file has email A31 then one or more email will
> be truncated by one character, which means data will be lost. For some
> people and some situations, data loss will be unacceptable. It seems
> obvious to me that the only way to prevent data loss would be to make
> the final merged string length equal to the longest of the widths
> encountered while merging.
>
> It sounds like you're suggesting that PSPP should just ignore this and
> continue but, if so, I don't understand what you mean by "ignore" and
> "continue"?
>
> Also, I've outlined why not saving the widest string for a variable is a
> problem; what harm would be created if the widest string was saved? Why
> is that a bad solution?
>
> -Alan
OK maybe I confuse width and content.
In the case of numeric variables, if I MATCH with a table file, and both 
files have a var TURNOVER, I know my current TURNOVER var will be left 
alone. The question whether the TURNOVER column in the table file may 
have better data than the one in my workfile is a different matter and 
of no interest to me here. In this case, with string variables, I would 
prefer to retain just my original variable width.
But I agree that with UPDATE in the case of strings the resulting string 
width could (should?) be set to the maximum width.
Friends :-)?
frans