Date: Sat, 12 Aug 2006 10:44:00 +0300 From: Giorgos Keramidas <keramida@ceid.upatras.gr> To: Julian Elischer <julian@elischer.org> Cc: current@freebsd.org Subject: Re: suggested addition to 'date' Message-ID: <20060812074400.GA9572@gothmog.pc> In-Reply-To: <44DD6CBC.9030309@elischer.org> References: <44DD4510.5070002@elischer.org> <20060812033607.GB80768@gothmog.pc> <44DD50FF.5040406@elischer.org> <20060812041535.GA82669@gothmog.pc> <44DD5992.5080409@elischer.org> <20060812045622.GA84354@gothmog.pc> <44DD6CBC.9030309@elischer.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On 2006-08-11 22:53, Julian Elischer <julian@elischer.org> wrote: >Giorgos Keramidas wrote: >>IIRC, stdio can buffer more than one line, so now that you mention it, >>maybe it is a good idea to flush at every '\n' character to make >>output appear every time there's a complete line ready. > > stdio will automatically flush pipe and terminal output at every \n. > the problem is if you are writing to a file. > If you get a signal it just calls _exit() which doesn't flush anything. > if it does an exit() it flushes the output so that would be ok. > signal handlers shouldn't call stdio as they are not async-safe, so making > a signal handler that calls fflush is not possible. > > I tried making the signal handler just set a variable that makes the > main loop quit, flush and exit, > but believe it or not, fgets() doesn't return from a signal. so you hit > ^C but it doesn't notice the flag that is set until > you then hit CR. hmm maybe if the signal handler closed file descriptor > 0....... This is getting too complex for my taste though. I don't see cat(1) doing signal trickery, so why should date(1) do these things? Perhaps it's not a good idea to 'bloat' date(1) so much...
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20060812074400.GA9572>