Stephen, I've built successfully on Solaris 8 SPARC (with & without gtk) two days ago and I've even sent the built reports. But I got an additional question (maybe it was answered before but I joined the beta fun not long ago): I'm currently on the gtk cvs branch of 21.2.44 and would like to know when & how the 21.2.45 changes will be merged in? RM> Please do builds of the gtk-xemacs sources --with-gtk=no. RM> Basically what I would like, as soon as possible, is lots of build RM> reports based on A small addition here which make the live a lot easier;-) Get the xemacs-sumo.tar.gz and extract it into a permantent location. RM> a cvs checkout of gtk-xemacs-21-2 ln -s <permanent/xemacs-packages> . RM> ./configure --with-gtk=no RM> make beta >beta.err 2>&1 RM> src/xemacs RM> M-x build-report Thanks, -- Marco Walther RM> It is important to do the explicit --with-gtk=no, even though that's RM> the default. The problem is there doesn't seem to be any other way to RM> recognize gtk-xemacs-21-2 sources in a build report, except for the RM> tester to put it in explicitly. RM> --with-gtk=yes reports are of course useful, but they are irrelevant RM> to whether GTK will be in the release since it will default to OFF. -BP> "Stephen J. Turnbull" <turnbull(a)sk.tsukuba.ac.jp&gt; writes: RM> There seem to be about 10 on Linux, one on OpenBSD, and no others. -BP> If anybody wants my insane makefile, let me know. Basic -BP> status right now is that for solaris, linux, and FreeBSD (and -BP> I did a few variant builds on cygwin the other day, but not -BP> my full suite), 48 different variants per platform get built -BP> at my house almost daily. -BP> gtk & x11 & tty, with/without union type, with/without MULE, -BP> pdump/unexec, and debug/release in all their variations build -BP> & run. RM> If you could run your insane make and submit them all to RM> xemacs-build-reports, it would be very useful. Then do it again with RM> warnings cranked far past any sane level, just for kicks. (I doubt I RM> would be able to interpret them, but it would be very useful in a RM> post-mortem kind of analysis.) RM> Doing the "submit the whole bunch" on a regular basis (say, every RM> couple of weeks) seems like a good idea to me. -BP> If there is an r21-2-45 tag that I could use as the target of -BP> the 'cvs diff', that would make life much easier. RM> There is as of a few hours ago. RM> -- RM> University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN RM> Institute of Policy and Planning Sciences Tel/fax: +81 (298) 53-5091 RM> _________________ _________________ _________________ _________________ RM> What are those straight lines for? "XEmacs rules." --

I'll present results to date in a separate mail. sjt> Please do builds of the gtk-xemacs sources --with-gtk=no. Oh, yeah. Do them on machines where GTK+ and GNOME are _not_ installed, too. That absolutely must work. Sandra> When do you need them by? Three weeks before the release date ;-) I apologize for another management inefficiency, but let's put it this way: do them as soon as you can, and watch xemacs-beta because I or Bill will announce when the urgency passes (basically, when Bill starts merging back to the release branch). It will still be useful even if later than that, but it probably can't affect the go/no-go decision. Sandra> what machines are most important to you? For Linux, anything non-Intel. Alpha and Sparc Linux, I guess are about the same priority as say Solaris (Solaris would be rather important, except that I know Bill will provide a lot of data there), but it's hard to judge. Any other Linuces would be lower priority. For non-Linux, whatever the most common platform for the OS is. Except that I'm getting a fair number of Sparc/Solaris builds recently, so Sparc/Solaris, like Intel/Linux is near zero priority--- if that's your main platform, by all means, submit build reports, it helps, but I would prefer any extra effort to go in other directions. Sandra> I can probably do a couple of builds on Monday at Red Hat, Sandra> but I need to know what the priority OSes are, because I Sandra> have too many available to me. NT native (I doubt you're set up for that, but others may be, and I should have been more specific about priorities). Cygwin. Irix. HPUX. NetBSD, OpenBSD. Then Solaris and FreeBSD, but Bill has done those and probably will report them very shortly. -- University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN Institute of Policy and Planning Sciences Tel/fax: +81 (298) 53-5091 _________________ _________________ _________________ _________________ What are those straight lines for? "XEmacs rules."

Bill, get me those build-reports for all your Insane Makefile platforms (restricted to --with-gtk=no if that helps). Successful builds --with-gtk=no on as many NT platforms as you can get would make me much happier. Cygwin is a must since I _know_ you have it. :-) Please address both me and xemacs-buildreports. I've only resubscribed to xemacs-buildreports in the last 24 hours and haven't gotten any yet. My procmail filter strips dupes so no problem there. I will review those as top priority, and give you the go to commit to the release branch, assuming nothing evil has shown up in the reports. -- University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN Institute of Policy and Planning Sciences Tel/fax: +81 (298) 53-5091 _________________ _________________ _________________ _________________ What are those straight lines for? "XEmacs rules."

Bill hasn't checked in with his reports yet. The three failures in i686-linux are all old, and due to the same problem in setting up the GTK device tag for the font/face specifiers, now corrected. Data is the monthly tarballs of xemacs-build-reports starting last August. There were no build reports in September. The results are basically the same if I just use the last few weeks, except that N is about 40. I will do this correctly later, but I think this version is of some interest for the list of platforms and frequency of reporting stats. N is the total number of build reports from August to date. GTK src is the number of reports based on src using GTK based on a crude heuristic. (see the Python script below.) gtk=no is the number of GTK source builds --with-gtk=no. OK is the number of successes, based on parsing the build report Subject for "success". rate = 100*OK/(gtk=no). Platform N GTK src gtk=no OK rate hppa1.1-hp-hpux11.00: 1 0 0 0 na hppa2.0-hp-hpux11.00: 1 0 0 0 na i386-debian-linux: 2 2 1 1 100% i386-pc-win32: 3 0 0 0 na i386-unknown-freebsd3.3: 6 0 0 0 na i386-unknown-freebsd3.4: 3 0 0 0 na i386-unknown-freebsd4.1: 7 0 0 0 na i386-unknown-freebsd5.0: 10 0 0 0 na i386-unknown-openbsd2.7: 2 0 0 0 na i386-unknown-openbsd2.8: 3 1 1 1 100% i586-pc-linux: 11 0 0 0 na i586-pc-sco3.2v5.0.2: 2 0 0 0 na i586-pc-win32: 4 0 0 0 na i686-pc-linux: 45 6 5 2 40% mips-sgi-irix6.2: 5 0 0 0 na powerpc-unknown-linux: 9 0 0 0 na sparc-sun-solaris2.5.1: 3 0 0 0 na sparc-sun-solaris2.6: 1 0 0 0 na sparc-sun-solaris2.7: 11 0 0 0 na sparc-sun-solaris2.8: 4 3 2 2 100% unknown: 2 0 0 0 na Totals (by hand) 135 12 9 6 67% #! /usr/bin/python # scaffolding = stuff that needs generalization dir = 'build-reports/' root = 'xemacs-build-reports.' stem = dir + root months = ['200008', '200010', '200011', '200012', '200101', '200102'] ## ## imports ## import re from string import strip, lower ## ## input functions ## def parseFile (filename): """Parse an xemacs-build-reports archive file. Assumes that a database (actually list of records) is in `db'.""" # slurp in file file = open(filename, 'r') lines = file.readlines() file.close() # process it for line in lines: # check for message start, initialize record if re.match('from ', line, re.I): inHeaders = 1 currentProperty = None record = { 'gtk' : '' } if re.search('gtk', line, re.I): record['gtk'] += line db.append(record) # should try to parse out an address continue if re.search('gtk', line, re.I): record['gtk'] += line # parse header lines if inHeaders: if re.match('^$', line): inHeaders = 0 continue m = re.match('([-_A-Za-z0-9]+)\s*:(.*)$', line) if m: currentHeader = lower(strip(m.group(1))) if not re.match(ignoredHeaderRE, currentHeader, re.I): record[currentHeader] = strip(m.group(2)) # print line, else: currentHeader = None elif currentHeader: if re.match('\s', line): # print currentHeader record[currentHeader] += ' ' + strip(line) else: # something evil happened record[currentHeader] += ' ' + strip(line) else: pass # parse body lines # assume each property is in a given section headed by a line matching # its keyword # sections terminated by empty lines elif not currentProperty: m = re.search('uname |configure |configured for ',line) if m: currentProperty = lower(strip(m.group(0))) if record.has_key(currentProperty): record[currentProperty] += '\n' + strip(line) else: record[currentProperty] = strip(line) elif re.match('uname|configure|configured for', currentProperty): if strip(line): record[currentProperty] += ' ' + strip(line) else: currentProperty = None else: print '***** currentProperty not handled: %s' % (currentProperty) ## ## functions to parse specific information from a record ## def parseSubject (record): """Parse status from Subject header of RECORD. Adds status field to RECORD.""" if record.has_key('subject'): m = re.search('success|failure',record['subject'],re.I) if m: record['status'] = lower(m.group(0)) else: record['status'] = 'unknown' def parseConfiguration (record): """Parse version, codename, and platform from configuration line of RECORD. Adds version, codename, and platform fields to RECORD.""" if record.has_key('configured for'): m = re.match('XEmacs (\\S*) \\\\?"(.*)\\\\?" configured for `(.*)\'', record['configured for'],re.I) if m: record['version'] = m.group(1) record['codename'] = m.group(2) record['platform'] = m.group(3) else: print 'Broken version: ' + record['configured for'] record['codename'] = 'unknown' record['version'] = 'unknown' record['platform'] = 'unknown' else: record['codename'] = 'unknown' record['version'] = 'unknown' record['platform'] = 'unknown' def parseConfigure (record): """Parse important switches out of the ./configure command line of RECORD. Adds appropriate switch fields to RECORD.""" if record.has_key('configure'): m = re.search('--with(out)?-gtk(=yes|=no)?',record['configure'],re.I) if m: if m.group(1): if m.group(2): record['with-gtk'] = 'confused' else: record['with-gtk'] = 'no' else: if m.group(2): record['with-gtk'] = lower(m.group(2)[1:]) else: record['with-gtk'] = 'yes' else: record['with-gtk'] = 'no' else: record['with-gtk'] = 'missing' ## ## reporting functions ## def gtkByPlatform (): """Check on --with-gtk=no testing by platform.""" gtkPlatforms = {} for record in db: if not gtkPlatforms.has_key(record['platform']): gtkPlatforms[record['platform']] = (0, 0, 0, 0) yes, no, withGTK, withoutGTK = gtkPlatforms[record['platform']] if record['gtk']: yes += 1 else: no += 1 if record['with-gtk'] == 'yes': withGTK += 1 elif record['with-gtk'] == 'no': withoutGTK += 1 gtkPlatforms[record['platform']] = (yes, no, withGTK, withoutGTK) print '%-30s %9s %9s %9s %9s' % ('Platform','GTK src','mainline','gtk=yes','gtk=no') keys = gtkPlatforms.keys() keys.sort() for platform in keys: print '%-30s' % (platform + ':'), print '%9d %9d %9d %9d' % gtkPlatforms[platform] def gtkIsNoSuccessRateByPlatform (): """Check on success rate of --with-gtk=no testing by platform.""" gtkPlatforms = {} for record in db: if not gtkPlatforms.has_key(record['platform']): gtkPlatforms[record['platform']] = (0, 0, 0, 0) n, GTKsrc, notGTK, ok = gtkPlatforms[record['platform']] n += 1 if record['gtk']: GTKsrc += 1 if record['with-gtk'] == 'no': notGTK += 1 if record['status'] == 'success': ok += 1 gtkPlatforms[record['platform']] = (n, GTKsrc, notGTK, ok) print '%-30s %3s %8s %8s %8s %6s' \ % ('Platform','N','GTK src','gtk=no','OK', 'rate') keys = gtkPlatforms.keys() keys.sort() for platform in keys: n, GTKsrc, notGTK, ok = gtkPlatforms[platform] print '%-30s' % (platform + ':'), print '%3d %8d %8d %8d' % (n, GTKsrc, notGTK, ok), if notGTK: print '%5s%%' % ((100*ok)/notGTK) else: print '%6s' % 'na' ## ## main program ## ignoredHeaderRE = 'received|to|cc|sender|x-.*|mime-.*|content-*' db = [] # create database for month in months: parseFile(stem + month) # process records for record in db: parseSubject (record) parseConfiguration (record) parseConfigure (record) #gtkByPlatform() gtkIsNoSuccessRateByPlatform() -- University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN Institute of Policy and Planning Sciences Tel/fax: +81 (298) 53-5091 _________________ _________________ _________________ _________________ What are those straight lines for? "XEmacs rules."