Hi Jonathan,
Hi Daniel. Good to hear from you. It's been a while.
In your platform[0], you state: I have a proven history of releasing software on time, on schedule. Project Xouvert, a stripped down version of the X11 source code, was released two times, six months apart. We didn't achieve many of our more ambitious goals, but we got a working release out the door on time, both times.
This did not in any way line up with my recollection of how Xouvert fared at all, so I challenged you about it on IRC:
Sorry I didn't get back to you sooner. I don't usually IRC from work, and just got home. Your perceptions are definitely of concern, so I'll address them here.
Contrast this with the Xouvert 0.1 announcement[2]:
Date: Sun, 7 Dec 2003 05:24:40 -0800
This release was either two months and six days, or one month and six
days, late; depends on how you look at it.
I didn't go digging through the archives before writing my platform. You may be right. Most of our announcements were done via our webpages. Since we lost our entire edit history for those when the repositories got wiped, I can't go through and show exactly how the December 7 release announcement matches up with "on time", but the whole team felt pretty proud of our schedule keeping.
I believe that by November 1 what we offered was the XFree86 sources just before the point where they changed the license, made available to the public through the arch revision control system. Since we hadn't made any changes, we didn't stamp it with a release number until we'd made some changes. Internally we called it the "developers release".
So, yes, we had shipping source code on time, but you are right, it wasn't a "release" as the outside world would consider it, involving modifications to the code, etc. Thanks for the trip down memory lane. I'm sorry if you feel misled; that wasn't my intent.
March's list traffic[10] sees your first post in months, in which you announce[11] that there has been a long radio silence, and that 'Xouvert at the moment is the XFree86 4.3 X server with Alan Cox's VIA drivers added'. There was no code behind this. Indeed, as you state later in the announcement:
I believe your statement on what Xouvert was at the time referred to plans, not code. You mention that you were probably moving to the commit repositories to fd.o; I do not recall this ever having happened, there is no 'xouvert' group on gabe.freedesktop.org[12], and no posts from you to [EMAIL PROTECTED] (the fd.o admin list).
Our activities on freedesktop.org all happened before gabe came up. I think we had an account on pdx. We used it as a mirror until the hard drive crash, but never uploaded the second release to it. Noone on the Xouvert team was ever made aware of the sitewranglers mailing list. Even the existence of lists.freedesktop.org is a fairly new thing.
The second release was real; we took the X.org sources, applied some patches, and ran it through a custom perl script that stripped the server code out from the rest of the CVS goop called X11. The release was downloadable from the Xouvert.org site.
While I was busy stamping out fires elsewhere, just before the second release, the webmaster had to deal with some issues on the server, and made some changes to the website, including moving things around. At that point, lacking repository histories, I threw in the towel; X.org was picking up steam, and although there was still a significant niche for Xouvert, everyone on the team, including myself, was moving on to other life obligations.
One of the (many) headaches that persuaded me not to revive Xouvert after the second release was the design of the arch revision control system; setting it up securely was not trivial at that point in its development. That may have changed; I respect the abilities of Tom Lord a lot. If I were to revive Xouvert today, I would use darcs. It is vastly easier to use and administer than arch. Darcs is what Xouvert needed from the beginning.
There were three follow-up questions (none with any real effect on the project, just musings about X, mainly; although an answer for 'what are you going to do now they've changed the licence?'
Xouvert dealt with the license change issue at its inception. Simply put, we grabbed the source code from just before the license change. Was there another license change since then?
Someone suggested rewriting the build system again.
I did. I planned to use the scsh scheme interpreter as a build system to replace imake, cpp, m4 and make in one felling swoop. Something it would still be nice to do. I don't believe in XML. It is a weak and sickly bastardization of SGML.
www.xouvert.org now states that the release slated for April 1, 2004 (sic; should be April 5), was 'the last release of Xouvert for now'[16] and that all the developers had moved on. However, there was never any announcement of: * a release, * a pending release, * a test release, * the archives being recreated, * anything other than 'XFree86 4.3 with Alan Cox's VIA patches'.
Yep, Xouvert certainly had problems.
a) the commit repositories weren't lost, and never recreated. b) testing was limited due to resource starvation at the end. c) yes, we fell down in announcing the release.
However, the pending release WAS announced long in advance, on the website. And the release was available on the website for a while, until things got reshuffled in the aftermath of a breakin.
I read your platform, and attempt to reconcile your statement from the first paragraph of this mail, with the history of Xouvert.
I hope I was able to clear matters up to your satisfaction.
Cheers!
Jonathan
--
It's not true unless it makes you laugh, but you don't understand it until it makes you weep.
Eukleia: Jonathan Walther Address: 12706 99 Ave, Surrey, BC V3V2P8 (Canada) Contact: 604-684-1319 (daytime) Contact: 604-582-9308 (morning and evening) Website: http://reactor-core.org/ Puritan: Purity of faith, Purity of doctrine Puritan: Sola Scriptura, Tota Scriptura
signature.asc
Description: Digital signature