[comp.sources.bugs] Inline - Known Bugs - from the Author

mcg@omepd (Steven McGeady) (09/25/87)

The following are known bugs in my inline posting.  I am working on
all of them.  Most of them are not fatal, and inline will remain useful
for most people until I can post fixes.

Current bugs:
Wed Sep 23 17:01:17 PDT 1987

	- the grep command line in the Makefile is wrong
	  [if you don't have RCS, just zero out the file vers.c and
	   leave it at that]

	- test 17, #3 fails - expressionized if stmts containing
	  returns are not handled correctly - old solution was better
	  [apologies - I applied the Patch too hastily]
	  [note: don't worry too much about this.  it only fails on
	   expressionized inlines with return statements inside if stmts]

	- reports are that the program fails on machines where
	  sizeof(char *) != sizeof(int), my apologies.  this one will
	  be hard to fix, since all I have are 32-bit linear address
	  space machines available (inline was tested on VAX,Sun,Pyramid,
	  and Gould)

	- unary ampersand (address-of) not understood as possible sideeffect
	  fix in either sideeffect() or isassigned()
	  [another rare one]

	- copy propagating global variables - see notes
	  [this is rare also, but hard to fix]

Thanks for help pointing these out to Anne Holler (uvacs!amh) (who has
written an inliner the 'right' way on her own), and to John Mackin
(john@basser.oz), and to Rick Richardson (pcrat!rick).

With regard to the copyright issues, please see my note in comp.sources.d.

S. McGeady