Does Roger do the flush as each file closes or does he wait and do one when J goes to immediate mode and flushes all writes? It's kind of like "commit" in data base. There needs to be a way to be assured that something is written to disk before an application proceeds. Right now it's not clear in J when that occurs.
On Sun, Apr 27, 2008 at 5:16 PM, Henry Rich <[EMAIL PROTECTED]> wrote: > Yes, that is clearly what is happening. I was just wondering > why the data gets flushed to disk before I can type the next > line into J. I would have expected that there would be a buzz > of disk activity while subsequent J sentences are being executed, > but it seems that something is preventing execution from continuing > until the files are synced out to the disk. I'm just curious. > > > When I said "Roger is in there doing something" I was reverting > to the way I talk to the highschoool kids. Rather than say, > "the interpreter executes your sentence" I just say, "Roger > executes your sentence from right to left", or "Roger can't make > any sense of what you just typed". They enjoy all the things > Roger does for them. > > Henry Rich > > > -----Original Message----- > > From: [EMAIL PROTECTED] > > [mailto:[EMAIL PROTECTED] On Behalf Of Eric Iverson > > Sent: Sunday, April 27, 2008 6:11 PM > > To: Programming forum > > Subject: Re: [Jprogramming] Does J flush files before > > returning toimmediatemode? > > > > In working on a project recently with many large files I have > > observed > > something that may be related to your experience. The host > > seems to do > > buffering that makes file io very quick. By quick I mean that > > the io appears > > finished to J so that J can continue but there is still > > processing to be > > done by the host to really get them on disk. But if there is > > enough io > > things eventually slow down so you see the real time to get > > the bytes to > > disk. Another way of thinking of this is that io for the > > first hundreds of > > magabytes is artificially fast, but after some point it > > starts to be a > > linear time per byte. > > > > ----- Original Message ----- > > From: "Henry Rich" <[EMAIL PROTECTED]> > > To: "'Programming forum'" <[email protected]> > > Sent: Sunday, April 27, 2008 3:55 PM > > Subject: [Jprogramming] Does J flush files before returning to > > immediatemode? > > > > > > > I'm curious about a behavior I have seen. > > > > > > I have a program that creates a few hundred files using > > 1!:2, and types > > > a message as it creates each one. Files are about 4MB each. > > > > > > It zips through them, and the messages fly by. But then, > > after all the > > > messages have been typed, there is a delay of maybe 10 > > seconds before > > > J will accept another command. The program is finished, > > but Roger is > > > in there doing something. > > > > > > Any idea what? > > > > > > Henry Rich > > > > > > > > > > > > > > > > > ---------------------------------------------------------------------- > > > For information about J forums see > > http://www.jsoftware.com/forums.htm > > ---------------------------------------------------------------------- > > For information about J forums see http://www.jsoftware.com/forums.htm > > > > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
