--- gforth/BUGS	1995/09/06 21:00:07	1.19
+++ gforth/BUGS	2003/02/03 08:59:28	1.52
@@ -1,26 +1,60 @@
-etags.fs crashes one of my applications (gs.fs). anton 12jan95
-
-not all aliases are in the etags file. Bug in etags.fs? anton 24jan95
--> wrote special tag generation for cross compilation. bernd 6sep95
-
-emacs often finds the wrong tag. anton 24jan95
--> emacs does not search for a complete word, but for parts.
--> Solution: add blanks in front and end of tag name (etags.fs)
--> and use own forth-find-tag. bernd 6sep95
-
 gforth.el: indentation does not work right on the first line of a
 buffer. anton 27jan95
--> Solved?
 
 Conditional compilation continues after the file ends. This is allowed
 by the standard (through an ambiguous condition), but the compiler
 should at least produce a warning.  anton 27jan95
 
-.1 is interpreted as floating-point number, not as double number or
-error. anton 4may95
--> All numbers ecvt converts are (by implementation) floating-point.
--> Should we add a check againgst . at the start of a number?
--> .1 isn't a double number; it's either an error or a FP number.
--> bernd 5sep95
-
-DOS and OS/2 don't like . at begin of a file (.gforth-history). bernd 6sep95py
\ No newline at end of file
+QUERY and TIB may behave differently than some may expect. TIB always
+points to the current SOURCE, and QUERY puts its result there.  anton
+28nov96
+
+open-path-file expands "./" into the sourcefilename. It should either
+not expand "./" or provide a mechanism that allows the application to
+determine what "./" should expand to. anton 16jun98
+
+gforth.el: Typing <ret> does not work when tabs separate words in a
+line, and there is a \-comment at the end of the line.  anton 19feb00
+
+SEE does not work when the control structure is too complex (e.g.,
+load http://www.complang.tuwien.ac.at/forth/pentomino.fs and then do
+SEE NEXT-PIECE). anton 5mar2000
+
+Include cannot handle lines longer than 255 characters.  anton 4sep00
+
+Errors happening during a LOAD do not report the offending word and
+its context (e.g., the 64-byte line). anton 8sep00
+
+Our ecvt routine apparently does not work correctly for Infs and NaNs.
+Try "ac_cv_func_ecvt=no ./configure; make" and then in Gforth: "1e 0e
+f/ f. 0e 0e f/ f.". anton 25sep00
+
+Our ecvt routine does not round correctly, e.g., 0.25->0.3. Marcel
+Hendrix 3oct00 <8rdcmd$j96$1@news.IAEhv.nl>
+
+Newline has only LF (instead of CRLF) in DOS.  Bruce Hoyt 25oct2000
+<39f7b14b$2@clear.net.nz>
+
+SIGPIPE can make Gforth hard to stop (e.g., 
+gforth -m 2M wordfreq.fs -e bye|head
+anton 30may01
+
+F. does not print trailing zeroes (e.g., "10 SET-PRECISION 125e f.")
+anton 31may01
+
+Block 0 does not work as it should: "0 block drop update save-buffers"
+does not write to the blocks file.  "0 block 1024 dump" seems to give
+the previous contents of the buffer.  Travis Bemann 10jul2001
+<3b4b4f57$0$42883$272ea4a1@news.execpc.com>
+
+When accessing a block beyond the end of the block file, the result is
+filled with spaces (this is also documented).  However, when accessing
+a previously unwritten block before the end of the block file, we will
+get a block full of zeroes on most (all?) OSs.  This inconsistency
+should be eliminated and the documentation fixed.  anton 14jul2001
+
+Path handling may have a bug:
+strace $FORTH 'fpath= .|~+/gforth s" /home/anton/threading-variations/gforth/mach32b.fs" include gforth/kernel/main.fs bye'
+tries to open gforth/kernel/gforth/kernel/aliases.fs, then
+gforth/gforth/kernel/aliases.fs, but not gforth/kernel/aliases.fs.
+anton 3 sep 2001