> Yes I was enquiring about objections from others > here. > If you have objections you mention it, which you > have done > anyway. So what is the problem with that ? > I am building FOX while having only FOX itself on > the build > machine. All Nevada Xwindows packages were removed.
I didn't remove them. I only renamed /usr/openwin to foo and made /usr/openwin a symlink pointing to /usr/X11. Plus I adjusted the symlinks in /usr/lib/. What's the big deal with getting it self-hosted? A few lib.so.N symlinks, adding missing libs to the system, proper build order. > So > this config is causing Xscreensaver to enable the > extrusion > plugins which fail without Freeglut and Gle. The > FOX packages > should be self-hosting. Mine are almost selfhosting. > I really do not understand what the problem is with > one merge > going before other either way. All you need to do > is a hg pull > and resolve the merge conflicts if any. > Alan has been periodically checking in stuff and I > have been > periodically merging while at the same time > testing and building. > What is the big problem with that ? And hg in fact Did he merge in very much? No. That's why it doesn't cause many problems. If he had changed 50% of the files, would it still have worked out so easily? > makes merging > a lot easier. Easier than with gdiff and SVR4-patch? How, not all those HUNK's? > This is not how opensource development works. One > does not > commit perfect huge codebase in one shot. Rather > you should be > committing often even if your changes are not fully > complete. This > also gives others to look at your code assist in > bug-fixing and > improving. It is an inclusive development model, > whereas what > you are doing is an exclusive one. That's right. And is compatible with my model. I only have a higher threshold value before I am willing to make the *initial* commit. That's the only difference. BTW, when did you last push something into FOX? > Have you posted your diffs at all ? You obviously never visited my website: http://www.martux.org/xorg/OLD/20070503__7.2.0/bin_pkgadd_packages/sparc-probe.patch http://www.martux.org/xorg/OLD/20070503__7.2.0/src/ - [DIR] 0__20070227/ 03-May-2007 15:54 - [DIR] 1__20070410/ 03-May-2007 15:54 - [DIR] 2__20070430moinakg__..> 03-May-2007 15:54 - [DIR] 3__20070430moinakg__..> 03-May-2007 15:54 - So then you also never tested my binary packages on SPARC? Only 3 IP addresses have downloaded the FOX1.0-SPARC tarball over the last 3 days ... :-( So you don't happen to be one of those 3 persons? > I have no clue > what changes > you made, I have posted a lot of info about them, to xwin-discuss. In contrast, why hadn't you posted or CC'ed more of your self-hosting discussions to xwin? I'm also a bit surprised by what you have done in that area already. > That sorted order looks more favourable to me. What > do you think? > > Please do not do one huge merge, it will cause > problems. The > FOX tree is currently being used for Project > Indiana and I have > been working to get a working resynced build. One > huge merge > will cause problems at the last momment. Have you Are you suggesting my whole effort was just for fun?? Why didn't you tell me earlier? > tested your > modified Xserver on x86 ? Sure, I had written about this. At first I had considered to integrate some basic logic into the make framework to apply patches depending on which ISA the build is run on/for. But then I thought about it: Mr. Coopersmith would certainly have done so, if he had wanted to, but he didn't. Probably for a good reason: Let's apply any and every patch, whether on SPARC or on x86, but lets keep the patches platform-clean by means of the conditional preprocessor directives. That's why my SPARC-patches do not interfere with x86. > Rather post your diffs somewhere, even if in the > form of a tarball > so that others can look at it. Then merge one > module at a time > not everything at once and not the Xserver changes > immediately > before it is tested on x86 - please post the > Xserver patchset and > I can then do the testing on x86. > What changes do you have for the apps stuff ? Just minor Makefile and pkgdefs related stuff. Plus a few build or download-target fixes. Just the few things that were required to add that stuff in ... Plus more or less mature pkgdefs, ISA-clean, with !search, license-header, etc. > Please post your > diffs first. I can hold back on merging my apps > stuff but I do not > understand what big problems it will cause. It may produce many HUNKS (if mercurial works like diff/patch, no idea). And one has to fix them one after another. Plus, I have really done something for making the pkgdefs look better, while there is still much work required. Just automatically merging stuff would break this, depending on how much you have extended yours (compared to the automatically generated ones from May). So okay. I won't come into your way. I won't sync in my stuff, if that's not wanted by Indiana. > R egards, > Moinak. Regards, Martin > > Regards, > > aplogogies (exclusively) to Moinak, > > > > Your friend Martin This message posted from opensolaris.org
