Re: utl_file and newline?

Simon Hall wrote:
>
> I've always found utl_file to be a pain for various reasons - line
> length limits, newlines etc and don't use it for anything more than log
> files now.
>
> As I understand it utl_file works by putting the text you're writing
> (with put or put_line) into a buffer (of very limited size). Then when
> you write a new line (either explicitly with new_line or implicitly at
> the end of a put_line call) the buffer is flushed to disk and a new line
> written. This flush/new line also happens when you close the file if
> there's anything left in the buffer.
>
> I had to produce some strictly formatted files recently on a windows
> system but with UNIX style line endings. I gave up with utl_file and
> wrote a java stored procedure instead. The downside is you've got to use
> 8i or later and need to install java support into your database.
>
> HTH,
>
> Simon
>
> Tim X wrote:
> >
> > Today I discovered something about utl_file and I just wanted to check
> > with others to verify I'm right - my conclusion seems correct, but I
> > find it difficult to believe this is the real behavior.
> >
> > I had a job where I need to create a file from data in the database
> > which has a specific format - lines must be EXACTLY 120 characters
> > long and the last line must NOT have a newline at the end.
> >
> > I initially thought utl_file would be fine for this as it has both
> > put, put_line and new_line procedures. However, it seems for some
> > difficult to understand logic that the fclose procedure will always
> > append a newline if one does not exist. I find this strange - why give
> > you control over the adding of newlines if the bloody package will
> > just go and stick one on the end wether you want ti or not!
> >
> > Has anyone else found this or is there something I'm missing? I've
> > tried some experiments and from what I can work out, fclose will
> > always add a newline at the end if one does not exist.
> >
> > I've now come to the conclusion the only way I will be able to create
> > this file is to write my data to a temporary table and then try
> > creating the file as I want it via spool from within sqlplus. Anyone
> > with a better solution?
> >
> > Tim
> >
> > --
> > Tim Cross
> > The e-mail address on this message is FALSE (obviously!). My real e-mail is
> > to a company in Australia called rapttech and my login is tcross - if you
> > really need to send mail, you should be able to work it out!

imho 32k is a pretty reasonable line width limit.

Cheers
Connor Received on Wed Feb 12 2003 - 02:33:16 CST