Re: Can we expect ISE Gui and makefile to produce identical bitfiles?
On May 22, 2:30*am, phil hays <philh...@dont.spam> wrote:
> As the .ise working file changes every run, and is binary to boot, it can
> not be an input into a stable and maintainable build process. So the
> solution I've used when using gnu make under Cygwin is to delete the
> whole result directory (bld) at the start of the build. There are other
> files in the result directory (and sub-directories under it) that can
> influence the build, and the only way that I'm aware of to get a
> consistent result is to start with a fresh directory.
Hi phil, thank you for your reply.
> One option for doing this would be to have the make file call a Project
> Navigator Tcl script (using xtclsh). This script would create a
> fresh .ise file every run, and could also be used to run from the GUI. I
> posted a script for this sometime ago, and will update it if desired.
This sounds useful - can you direct me towards a recent version of
this please?
> This is because the ISE flow seems to read the UCF file into a data base
> first, and then applies the constraints later.
However the ISE flow is using the '-i' flag, which is supposed to
ignore constraints... I only get the same behaviour from the command
line if I ditch the -i flag and use -uc instead, to include
constraints.
> The .ise file has lots of date and time information. The solution to this
> is to think of the .ise file as a working file, rather than a project
> file and to delete it at the start of any build script.
In conjunction with your earlier comment, this makes sense. However I
understand if -intstyle ise is *not* used, there's no dependency on
the ise file whatsoever. I'd prefer this at build time, although we
would like to automatically create an ise file for local use inside
the GUI.
> To difference the .bit files, the header needs to be ignored. To make
> this automatic, I've written a little difference utility using Tcl. Would
> this be of interest?
Yes, this would be useful please.
Regards,
-- David.
|