- 2010.02.26 14:30 "Re: [Tiff] Patch for tif_unix.c", by Edward Lam
-
2010.02.26 17:05 "Re: [Tiff] Patch for tif_unix.c", by Bob Friesenhahn
- 2010.02.26 17:43 "Re: [Tiff] Patch for tif_unix.c", by Edward Lam
-
2010.03.02 08:14 "Re: [Tiff] Patch for tif_unix.c", by Eric Doenges
- 2010.03.02 14:11 "Re: [Tiff] Patch for tif_unix.c", by Toby Thain
- 2010.03.02 14:21 "Re: [Tiff] Patch for tif_unix.c", by Edward Lam
2010.03.02 16:55 "Re: [Tiff] Patch for tif_unix.c", by Toby Thain
On 2-Mar-10, at 11:14 AM, Bob Friesenhahn wrote:
On 2-Mar-10, at 9:53 AM, Scott Ribe wrote:
Unless I'm missing something.
Yes, generally the correct response to EINTR is to retry, because it is simply the result of the kernel dealing with a signal, and not any kind of actual I/O failure.
Right, but what was the intention of the signal?
That is the crux of the matter. An application which handles signals would need to have a way to abort all I/O loops if it wants to quit (e.g. if the user hit ^C). Otherwise the program hangs and the user becomes annoyed.
That was my line of thinking.
--Toby
Libtiff does not include such smarts. Everything becomes more challenging if the application is threaded.
> bfriesen@simple.dallas.tx.us, http://www.simplesystems.org/users/
> bfriesen/
GraphicsMagick Maintainer, http://www.GraphicsMagick.org/