home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.lang.fortran
- Path: sparky!uunet!rational.com!voltaire!davidm
- From: davidm@voltaire.Rational.COM (David Moore)
- Subject: Re: unformatted output: << SUMMARY >>
- Message-ID: <davidm.721508082@voltaire>
- Sender: news@rational.com
- Organization: Rational
- References: <1992Nov4.160112.26525@chpc.utexas.edu> <1992Nov5.111303.17921@uts.uni-c.dk> <1992Nov11.130931.2171@jyu.fi>
- Date: Wed, 11 Nov 1992 18:54:42 GMT
- Lines: 29
-
- sakkinen@jyu.fi (Markku Sakkinen) writes:
-
- >In article <1992Nov5.111303.17921@uts.uni-c.dk> ohnielse@oersted.ltf.dth.dk writes:
- >>In article 26525@chpc.utexas.edu, michael@chpc.utexas.edu (Michael Lemke) writes:
- >>>Doesn't have the record *number* to be stored at some place as well?
- >>>Nothing stops you from doing
- >>>
- >>> open( ,.... access='direct', .... )
- >>> write( ..., rec=777 ) data
- >>> write( ..., rec=32 ) mode_data
- >>> close( ... )
- >>>
- >>>which gives you just two records of data and lots of `holes'. Or am I
- >>>missing something?
-
-
- This is often true. However, some Fortran systems hook into some
- sort of "Record Manager" interface, in which case you may be able to
- control the organisation of your file with system directives, or with
- extended directives in your "open", or else with calls to
- system routines with lots of "$"s in their names. You may be able
- to convince the system to give you any of 57 varieties of indexed
- file, in which the record number is stored as an index with a pointer
- to the data.
-
- This will avoid allocating space for unused records. It may also allow
- variable length records. Need I also remark that its not portable?
-
-
-