In some of these cases, I think that PSPP could issue fewer warnings.
For example, when GET DATA finds bad data, it is probably useful to
issue one warning, but it might not be valuable to issue 100 warnings.
Perhaps it should simply summarize ("found 325 additional invalid
observations") or the details could be put into notes instead of
warnings. Would that make better sense? I don't know what SPSS does in
similar circumstances.
When PSPP actually does issue warnings for good reasons, honoring
MXWARNS is a matter of compatibility with SPSS. It's possible to turn
the MXWARNS feature off, I believe, with SET.
On Fri, Nov 15, 2019 at 11:51 PM Frans Houweling <address@hidden> wrote:
Hi Ben,
GET DATA on csv files is the main source, like when missing values are coded "NA" or
":".
Plus I do make the ehm occasional mistake - but as what I do wrong in the first
record I do wrong in all the following records too, I always reach MXWARNS. I
can see some justification for halting in the case of vector index out of
bounds, but other errors like divide by zero cause no harm at all, like the GET
DATA case.
So if I could decide, I would vote for continue syntax processing.
Thanks
frans
On 11/16/19 4:35 AM, Ben Pfaff wrote:
I'm a little surprised you're receiving so many warnings. Is that a problem of
its own? What's the underlying cause?
On Fri, Nov 15, 2019, 12:36 PM Frans Houweling <address@hidden> wrote:
Correction: syntax processing does stop - only not in the first
procedure. Bad!
frans
On 11/15/19 8:47 PM, Frans Houweling wrote:
Hi,
I often see this message
"note: Warnings (n+1) exceed limit (n). Syntax processing will be
halted."
Luckily, syntax processing seems to go on anyway. I think the message
should just say "No more warnings will be issued" and syntax
processing should not be halted.
I am aware SET MXWARNS=0 suppresses all warnings - but I like to see
at least one (or possibly one per type).
Greetings
frans