335180 - Remove win32.order, mozilla-bin.order, --enable-reorder, and associated code

Status: RESOLVED FIXED

(Firefox Build System :: General, enhancement)

Description

[Christian :Biesinger (don't email me, ping me on IRC)]

I want to remove the order stuff in the tree. Reasons:
- The GCC --enable-reorder option is not useful:
  - It only affects the suite binary (i.e. at best, useful for a static seamonkey build)
  - mozilla-bin.order uses GCC 2.x name mangling
- win32.order files are only used if the environment variable MOZ_COVERAGE is set. 
  This is not the case on either the Firefox release tinderbox (http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1145826600.1074.gz&fulltext=1) nor the SeaMonkey one (http://tinderbox.mozilla.org/showlog.cgi?log=SeaMonkey/1145823060.12133.gz&fulltext=1)

  So clearly nothing uses them, and these files haven't been updated in ages (but should have been, as the functions are changing all the time)

This code was added in bug 65845.

If you can see a reason why this shouldn't be done, please say so :-)

Comment 1

[Christian :Biesinger (don't email me, ping me on IRC)]

Attachment: like this

in the interest of a shorter patch, I abbreviated the deletes here

Other than *.order, this deletes:
- config/trace.cpp
- config/*.inc (seems to have been used by the nmake build system only)
- config/mklinkscript.pl

Comment 2

[Darin Fisher]

It seems to me like it might be worth it to resurrect the reorder option instead of killing it outright, no?

Comment 3

[Mike Shaver (:shaver emeritus)]

We can resurrect these from CVS if/when someone decides to set us up for this bomb again, I think.  I prefer removing clutter (#if 0, unbuilt files, etc.) in the presence of even the minimal version management tools that CVS provides. :)

Comment 4

[Christian :Biesinger (don't email me, ping me on IRC)]

fixed on trunk, except for js/src/win32.order - I have no permission to check into spidermonkey.

Comment 5

[Enrico Weigelt]

I actually didn't yet understand, what this should be good for at all.
Having reducing complexity in mind, I think its an good idea to remove it.