>>Peter,
>>First of all bold is not your fault :) It was a forgotten closing tag in message you replied first.
>>You can find all the details of RegExp object in vbscript5.chm. I was disappointed Ed didn't also write appending part in WSH :)
>>Now your situation is clearer and also why it runs slow. substr() is slow to operate on strings whether or not it has been enhanced in VFP6. Why don't you try lowlevel once. Or get into a cursor and do one replace all. Sorry I insist. Lowlevel and FLL versions can parse a file word-by-word in approx. 1 sec per Mb (500Mhz). I don't think substr() could catch it even in VFP6. If it can I promise to keep my mouth shut.
>>Best regards.
>>Cetin
>
>
>Cetin,
>Thank´s for your continuing interest. I´ve tried to avoid having to explain the whole scenario, because it´s quite complicated, and I didn´t want to bother you people with it -but, you asked for it:-)so here we go...
>
>The file I´m trying to read is off a NT performance monitoring system called NTSMF and contains:
>
>1. a bunch of so-called discovery or HEADER-records, which essentially are records that tell you about the DATA records later on in the file. Without the presence of these records, the data records are just so much gibberish.
>
>2. the DATA records, consisting of a
>RECORDHEADER wich tells you what is being measured (i.e. Memory,Cache a.s.o)
>
>the POSITION of the first data field in any given line in the file
>
>the NUMBER of DATAFIELDS that should be read
>
>and finally, in conjunction with the discovery records, you need to figure out which fields to NOT to read.
>
>So this is the reason, why I´m going through the file the way I am.
>
>I am sort of turning the record around and making it usable for a VFP/SQL server database.
>Sorry, but in this instance a cursor is not the way to go.
>
>Finally,which lowlevel and FLL functions are you referring to?
>BTW, thank´s for the additional info on WHS!

Peter,
If you think 'a cursor is not the way to go' I can't make you believe it's a way to go. Just think any file with a linelength of max 64770 bytes + CRLF (w/o memo usage) could be read into a cursor. RECORDHEADER line(s) would provide the info what the structure would be for another append cursor. IOW I say read file as if it was all 255*c(254) fields first (no loss since any char field in fox could hold unprintable chars like chr(0) and ch(255) too). If you provided RECORDHEADER + 10 lines as sampling in text format life would be easier for us to analyze :)
Lowlevel fuctions are fcreate(), fopen(), fread(), fgets(), fputs() etc. FLL was a special one I wrote for parsing text files to words-tokens.
Cetin