Date:         Sat, 23 Aug 2003 08:41:16 -0400
Reply-To:     jsl <nospam@NOSPAM.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         jsl <nospam@NOSPAM.COM>
Organization: University of Georgia
Subject:      Re: Trouble reading large CSV files with about 80 variables

Thanks.

And, I understand your suggestion about e-mail. However, I avoid putting my
e-mail anywhere I don't have to, due to way too much spam.  Hackers are
smart enough to scan text of posts as well.  I'd rather just post and read
posts directly on the board.  There's no need for my e-mail address.

Thanks,
Jim

"Ian Whitlock" <WHITLOI1@WESTAT.COM> wrote in message
news:9B501B3774931C469BCCCC021BE537223D49BA@remailnt2-re01.westat.com...
> Jim,
>
> Best guess is that you are not making the LRECL parameter large enough and
> probably not using TRUNCOVER or MISSOVER on the INFILE statement.  Bad
> guess, show code with example of a problem record.
>
> I do not mind people who want to mess up the reply field to avoid spam,
but
> it does seem reasonable to then put the correct address in the text for an
> easy return.
>
> IanWhitlock@westat.com
>
> -----Original Message-----
> From: jsl [mailto:nospam@NOSPAM.COM]
> Sent: Friday, August 22, 2003 4:25 PM
> To: SAS-L@LISTSERV.UGA.EDU
> Subject: Trouble reading large CSV files with about 80 variables
>
>
> I've been reading .csv files for years with no trouble, but lately I've
> started to have trouble.  Hopefully the following makes some sense and
> someone can offer guidance.
>
> I have a CSV file with about 50,000 records and about 80 variables, many
> numeric, many date and many character.  I can't figure out what I'm doing
> wrong anymore.  Many of the variables show up as missing when they should
be
> text variables, many show up as the wrong variable (as if it wrapped),
etc.
> I have defined the informat for every non-numeric variable.  I created the
> .csv file using Excel.  I also tried a proc import to get the data from
> Excel, but it doesn't want to read a couple of my character variables as
> character variables -- i.e., it insists on treating it as a numeric field
> and generates missing values for those records that have character data in
> the field.
>
> Anything jump out at you?
>
> Thanks,
>
> Jim