many lines are discarded.
Are you suggesting that I list-loop an initial pass against the entire
file creating an array, then parse the array line-by-line?
Mmmmm.
Dick
On Jun 15, 2004, at 5:56 PM, Barney Boisvert wrote:
> I can't help you with the CFFILE issue (though you could probably
> implement
> it in java without great difficulty), but it seems that converting
> the file
> to an array would get around your issues with line pointers. You can
> add or
> remove lines in the middle, keep track of where you are in the file,
> and do
> arbitrary reads and skips at any point. I don't know what the
> overhead of
> that would be, but I'd imagine it wouldn't be any slower than the
> loop.
>
> Cheers,
> barneyb
>
> > -----Original Message-----
> > From: Dick Applebaum [mailto:[EMAIL PROTECTED]
> > Sent: Tuesday, June 15, 2004 5:45 PM
> > To: CF-Talk
> > Subject: Re: Out of Memory Error Transforming large XML file.
> >
> > I finally chose to parse it myself, recoding existing Perl routines
> > into CFMX scripts.
> >
> > So far it looks promising -- Most Perl structure, syntax & RegExps
> > convert quite easily to CFScript.
> >
> > There are still a few disadvantages to CFMX over Perl
> >
> > 1) CF lacks a file readLn (readLine), so you must read in the entire
> > file into memory. You then can approximate a Perl chomp
> > (readLn) by CF
> > list-looping over the file with a newline delimiter. Note: Using
> CF
> > list-looping presents the next line with each iteration and performs
> > well (ListGetAt() is too slow to be acceptable).
> >
> > 2) Perl maintains a pointer to the current line position in a
> > file --
> > this means you can chomp (readLn) the file in subroutines,
> > then return
> > to the caller with the current line positioned where the subroutine
> > left it. CF list-looping has no similar construct, nor the
> > ability to
> > list-loop from-to -- this means that the logic of the main
> > routine gets
> > more complex and there is more main-subroutine overhead.
> >
> > For example the xml file is divided into two major sections: tracks
> > followed by playlists.
> >
> > The Perl program chomps till it finds the first track
> > ----passes control to a get_tracks subroutine
> > ------subroutine chomps tracks until first play list, then returns
> > Main program passes control to get_playlists subroutine
> > ------subroutine chomps playlists until EOF, then returns
> > Main program continues
> >
> > In the CF program
> >
> > Main Program loops over the list line by line.
> > if found first track and not first playlist
> > ---call get_tracks sub to process this track (line) only & return
> > if found first playlist
> > ---call get_playlist sub to process this playlist (line) only &
> return
> > Main program continues
> >
> > As you can see, there is additional testing and call
> > overhead for each
> > line in the CF solution -- all because CF can't readLn a file nor
> > manipulate the list-loop position in a subroutine.
> >
> > For the latter I would like to see CFML improved to include:
> > 1) an implied list pointer for each list loop:
> > 1) , ListGetFirst, ListGetLast, ListGetAt functions that
> > reposition the
> > list pointer
> >
> > Also cf list-loop with from and to parameters
> >
> > Also CFFile readln
> >
> > Dick
>
[Todays Threads] [This Message] [Subscription] [Fast Unsubscribe] [User Settings]
