C Programming Language :: Signalhandler and "-ansi -pedantic -Wall"


	Signalhandler and "-ansi -pedantic -Wall" Hi, I've got a problem. I wrote a Program with a Signalhandler Routine. If i use the compile flags "-ansi -pedantic -Wall", "(void) signal(SIGCHLD,Handler)" catches the first signal but ignores the others. The installed handler seems to be gone after the first run. If I remove the "-ansi -pedantic -Wall" flags, everything works just fine. But why? thx a lot for your answers in advance ! On Thu, 31 May 2007 23:01:02 +0200, Simon Klein <a@me.at> wrote: >Hi, >I've got a problem. I wrote a Program with a Signalhandler Routine. If i use >the compile flags "-ansi -pedantic -Wall", "(void) signal(SIGCHLD,Handler)" >catches the first signal but ignores the others. The installed handler >seems to be gone after the first run. >If I remove the "-ansi -pedantic -Wall" flags, everything works just fine. >But why? >thx a lot for your answers in advance ! Off topic here. You should ask in a forum for the platform you're working on, comp.unix.programmer, at a guess. As another guess, you should probably be re-issuing the signal() in the handler. OTOH, you probably should really be using sigaction() <g>. Go ask the experts in comp.unix.programmer. -- Al Balmer Sun City, AZ - Hide quoted text - Al Balmer wrote: > On Thu, 31 May 2007 23:01:02 +0200, Simon Klein <a@me.at> wrote: >>Hi, >>I've got a problem. I wrote a Program with a Signalhandler Routine. If i >>use the compile flags "-ansi -pedantic -Wall", "(void) >>signal(SIGCHLD,Handler)" catches the first signal but ignores the others. >>The installed handler seems to be gone after the first run. >>If I remove the "-ansi -pedantic -Wall" flags, everything works just fine. >>But why? >>thx a lot for your answers in advance ! > Off topic here. You should ask in a forum for the platform you're > working on, comp.unix.programmer, at a guess. As another guess, you > should probably be re-issuing the signal() in the handler. OTOH, you > probably should really be using sigaction() <g>. Go ask the experts in > comp.unix.programmer. thx and sorry for the offtopic post In article <465f399c$0$25625$91cee@newsreader02.highway.telekom.at>, Simon Klein <a@me.at> wrote: >I've got a problem. I wrote a Program with a Signalhandler Routine. If i use >the compile flags "-ansi -pedantic -Wall", "(void) signal(SIGCHLD,Handler)" Compile flags are not specified by the C standard, so to find out more about which compile flags you should use and what their effects are, you should consult with a newsgroup that deals with your toolchain. It appears that you are using gcc, so one of the gnu.* newsgroups would seem most appropriate. >catches the first signal but ignores the others. The installed handler >seems to be gone after the first run. >If I remove the "-ansi -pedantic -Wall" flags, everything works just fine. >But why? <OT> I dunno. But signal semantics have changed over the years. One of the older common signal() semantics is that signals do not automatically rearm; if you are using such a system, then you have to reissue the signal() before leaving the signal handler. As to why it happens with those flags but not without them -- well, possibly not having to rearm is an extension and -ansi is requesting that extensions not be used. Or there might be some other reason for it; ask in an appropriate newsgroup. -- All is vanity. -- Ecclesiastes On May 31, 5:26 pm, Al Balmer <albal@att.net> wrote: - Hide quoted text - > On Thu, 31 May 2007 23:01:02 +0200, Simon Klein <a@me.at> wrote: > >Hi, > >I've got a problem. I wrote a Program with a Signalhandler Routine. If i use > >the compile flags "-ansi -pedantic -Wall", "(void) signal(SIGCHLD,Handler)" > >catches the first signal but ignores the others. The installed handler > >seems to be gone after the first run. > >If I remove the "-ansi -pedantic -Wall" flags, everything works just fine. > >But why? > >thx a lot for your answers in advance ! > Off topic here. You should ask in a forum for the platform you're > working on, comp.unix.programmer, at a guess. As another guess, you > should probably be re-issuing the signal() in the handler. OTOH, you > probably should really be using sigaction() <g>. Go ask the experts in > comp.unix.programmer. signal() is ansi C, isn't it? By the way, the manpage seems to indicate that if you compile with different defines, different semantics are used. I suppose removing the -ansi changes the defines and allows the use of another semantic than the one of ansi C. In article <1180647399.010236.59@p77g2000hsh.googlegroups.com>, Quentin Godfroy <quentin_godf@hotmail.com> wrote: >On May 31, 5:26 pm, Al Balmer <albal@att.net> wrote: >> On Thu, 31 May 2007 23:01:02 +0200, Simon Klein <a@me.at> wrote: >> >I've got a problem. I wrote a Program with a Signalhandler Routine. If i use >> >the compile flags "-ansi -pedantic -Wall", "(void) signal(SIGCHLD,Handler)" >signal() is ansi C, isn't it? SIGCHLD is not. -- "It is important to remember that when it comes to law, computers never make copies, only human beings make copies. Computers are given commands, not permission. Only people can be given permission." -- Brad Templeton On May 31, 5:39 pm, rober@ibd.nrc-cnrc.gc.ca (Walter Roberson) wrote: > >On May 31, 5:26 pm, Al Balmer <albal@att.net> wrote: > >> On Thu, 31 May 2007 23:01:02 +0200, Simon Klein <a@me.at> wrote: > >> >I've got a problem. I wrote a Program with a Signalhandler Routine. If i use > >> >the compile flags "-ansi -pedantic -Wall", "(void) signal(SIGCHLD,Handler)" > >signal() is ansi C, isn't it? > SIGCHLD is not. Whoops. Sorry for the mess. On 31 May 2007 14:36:39 -0700, Quentin Godfroy - Hide quoted text - <quentin_godf@hotmail.com> wrote: >On May 31, 5:26 pm, Al Balmer <albal@att.net> wrote: >> On Thu, 31 May 2007 23:01:02 +0200, Simon Klein <a@me.at> wrote: >> >Hi, >> >I've got a problem. I wrote a Program with a Signalhandler Routine. If i use >> >the compile flags "-ansi -pedantic -Wall", "(void) signal(SIGCHLD,Handler)" >> >catches the first signal but ignores the others. The installed handler >> >seems to be gone after the first run. >> >If I remove the "-ansi -pedantic -Wall" flags, everything works just fine. >> >But why? >> >thx a lot for your answers in advance ! >> Off topic here. You should ask in a forum for the platform you're >> working on, comp.unix.programmer, at a guess. As another guess, you >> should probably be re-issuing the signal() in the handler. OTOH, you >> probably should really be using sigaction() <g>. Go ask the experts in >> comp.unix.programmer. >signal() is ansi C, isn't it? Yes, but the re-arming behavior is implementation defined. I don't know whether POSIX defines it. >By the way, the manpage seems to indicate that if you compile with >different defines, different semantics are used. I suppose removing >the -ansi changes the defines and allows the use of another semantic >than the one of ansi C. -- Al Balmer Sun City, AZ Al Balmer <albal@att.net> writes: > On Thu, 31 May 2007 23:01:02 +0200, Simon Klein <a@me.at> wrote: >>I've got a problem. I wrote a Program with a Signalhandler Routine. If i use >>the compile flags "-ansi -pedantic -Wall", "(void) signal(SIGCHLD,Handler)" >>catches the first signal but ignores the others. The installed handler >>seems to be gone after the first run. >>If I remove the "-ansi -pedantic -Wall" flags, everything works just fine. >>But why? > Off topic here. ... because the C standard doesn't specify what happens to the handling of a signal when a signal handler is invoked. Both behaviors you're seeing are consistent with what the C standard specifies. (It's deliberately vague so that varying existing behaviors are permitted.) > You should ask in a forum for the platform you're > working on, comp.unix.programmer, at a guess. As another guess, you > should probably be re-issuing the signal() in the handler. OTOH, you > probably should really be using sigaction() <g>. Go ask the experts in > comp.unix.programmer. Yes, but first read your system's documentation for the signal function. <OT>On Unix-like systems, there may be multiple man pages for "signal"; "man man" if you don't already know how to find the one you need.</OT> -- Keith Thompson (The_Other_Keith) k@mib.org <http://www.ghoti.net/~kst> San Diego Supercomputer Center <*> <http://users.sdsc.edu/~kst> "We must do something. This is something. Therefore, we must do this." -- Antony Jay and Jonathan Lynn, "Yes Minister"