From: jeffdbREMOVETHIS AT netzone DOT com (Mikey) Subject: Re: rcs == whirling blades of death in B19 29 Mar 1998 20:47:26 -0800 Message-ID: <351a0b8b.49079222.cygnus.gnu-win32@smtp.netzone.com> References: <2 DOT 2 DOT 32 DOT 19980323151036 DOT 00dbcff8 AT agames DOT com> Reply-To: jeffdbREMOVETHIS AT netzone DOT com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit To: "David O'Riva" , gnu-win32 AT cygnus DOT com This is how I fixed the problem for B17.1 use this in addition to the other rcs patch posted to the list. --- conf.sh.orig Mon Mar 17 18:15:23 1997 +++ conf.sh Mon Mar 17 18:22:37 1997 @@ -1890,10 +1890,11 @@ echo "$a#undef EXIT_FAILURE $z/* Uncomment this if EXIT_FAILURE is broken. */" : configuring large_memory -case "$has_map_fd$has_mmap" in -*1*) l=1;; -*) l=0 -esac +#case "$has_map_fd$has_mmap" in +#*1*) l=1;; +#*) l=0 +#esac +l=1 #force large_memory to overcome bug echo "#define large_memory $l /* Can main memory hold entire RCS files? */" $ech >&3 "$0: configuring LONG_MAX $dots" --- rcstest.orig Mon Mar 17 19:14:23 1997 +++ rcstest Mon Mar 17 19:14:51 1997 @@ -173,7 +173,7 @@ case $USER in ?*) me=$USER;; *) - me=`who am i` || exit 2 + me=`whoami` || exit 2 me=`echo "$me" | sed -e 's/ .*//' -e 's/.*!//'` case $me in '') echo >&2 "$0: cannot deduce user name"; exit 2 On Mon, 23 Mar 1998 07:10:36 -0800, you wrote: >Running B19 w/coolview (March 15 version) on NT 4.0 SP3: > CYGWIN32=title binmode notty > all mounts are text=binary > >rcs is eating our files! I installed rcs-5.7, and applied wolfgang's patch >to get it to configure and build. Everything looked great, until it >suddenly and inexplicably munched a revision storage file. This is a Bad >Thing. > >Experimentation revealed the following: > > * The problem only happens when a stored revision is longer than 1K. > * The problem is very sporadic > * When it happens, the stored "text" of a revision is truncated to exactly > 1024 bytes and it's position in the revision tree is changed!!! > >Further investigation seems to indicate that something is going wrong in the >guts of fastcopy(). What's strange is that everything starts to work >correctly when the has_mmap flag is forced ON in conf.h, even though >configure blows up when trying to determine the status of mmaping on the >system. There was a message earlier on in the list about configure blowing >up when trying to check a situation which is impossible under 95/NT, which >prompted my trying this - is this a reasonable solution? > >We are trying to find and solve this problem here, but if any of you have >run into it before (or can categorize it faster than we can!), some advice >would sure be appreciated. > >thanks! >-dave ________ _______ ____ >---------------------------------------------/_ \ \/| ___ \ \- - >David O'Riva - Staff Programmer /| |\ \| ____/ /\ \ > oriva AT agames DOT com /| |/ /| |\/\ \/ /\ > "2^168 to 1 against and counting..." __|_____o/|__o\ /\____o\ > - - ---/________\/____\--/____\-- - > >- >For help on using this list (especially unsubscribing), send a message to >"gnu-win32-request AT cygnus DOT com" with one line of text: "help". ===================================================== Linux a platform built by, and for users, standing on the firm legs of reliability, and speed. Microsoft Windows, a platform without a leg to stand on. (jeffdbREMOVETHIS AT netzone DOT com) delete REMOVETHIS from the above to reply Mikey - For help on using this list (especially unsubscribing), send a message to "gnu-win32-request AT cygnus DOT com" with one line of text: "help".