Date: Sun, 14 Sep 1997 19:57:33 +0300 (IDT) From: Eli Zaretskii To: DJ Delorie cc: djgpp-workers AT delorie DOT com Subject: Alpha snapshot 970831 Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Precedence: bulk Here's some comments on that snapshot. I'm sending the patches in separate messages. 1) go32-v2 should have its version bumped to 2.02 in the blurb it prints when invoked with no arguments. 2) src/docs/wc202.txi sometimes use @code{func()}, which I think is incorrect, because it looks like a call to func with no arguments. @code{func} is how it should be written, as it is in some of the places and in the rest of the docs. 3) I think we should change the library functions which call dpmi_allocate_dos_memory to call the DOS function via __dpmi_int, and the same for free_dos_memory, because programs that call the free function will crash on NT if they use the DPMI call. There are only a couple or so such functions, so this shouldn't be too hard. 4) README.1ST is the old version. I thought I saw somebody posting an updated version to the news group (after DJ has provoked him); why not use it? Also, the list of files in README doesn't include the new zoneinfo zip. 5) Are we going to add the new headers and functions as per the ANSI C Addendum (, , )? 6) What do people think about making `printf' work (through a BIOS call) in graphics modes? Borland obviously does this, and it helps in debugging graphics programs. It is a matter of simple change in `doprint', but I wanted to hear opinions before submitting diffs. 7) When both "foo" and "foo.exe" exist in the current directory, ``spawnv("foo")'' invokes "foo" rather than "foo.exe". This is inconsistent with `spawnvp' which finds "foo.exe" first. Is this intentional or a bug? (One situation when this does matter is if you have stubedited foo.exe to have non-default stack or transfer buffer size.)