Stuff Michael Meeks is doing
|
|
This is my (in)activity log. You might like to visit my employer
Novell which is an amazing company, and also
Dell who in days of yore provided me with a
free laptop for Gnome development / conferences.
Also if you have the time to read this sort of stuff you could enlighten
yourself by going to Unraveling Wittgenstein's net or if
you are feeling objectionable perhaps here.
Older items:
2008: (
J
F
M
A
M
J
J
A
),
2007: (
J
F
M
A
M
J
J
A
S
O
N
D
),
2006,
2005,
2004,
2003,
2002,
2001,
2000,
1999,
legacy html
- Up early, to battle - call with Sabine, Tor,
Michael, poked Dick, Miguel - most helpful. More research.
Lydia over for lunch.
- Filed Patrick's pet bug, more mail churn,
conf call, one-on-one call, dinner with Tony & Janice,
Cell group, bed.
- Up early, feeling dead, to work. Tim Janik
recommended that I make the fact that
Novell are still hiring for
one position on OO.o - and perhaps other roles. Candidates
should be excellent programmers, C++, XML, octopus-wrestling
etc. useful but not required. Restrained my blink-tag urge,
mail me ASAP if you're interested.
- Poked at one of those irritating a11y / threading
issues, team meeting, misc research, call Nat, Simon, poked
Miguel, and still no hacking done, more research. Nearly got
on top of mail.
- Up late, played with babes a little, lunch,
to work. Great news: today Hubert Figuiere
joined the Novell OO.o team to work on OpenXML support -
initially for Impress.
- Dug at the bottom of the mail-mountain, an
up-hill experience for sure. Conf call late, bed early.
- Up lateish, breakfast, hacked at my talk, off
to the venue - saw several fine talks, gave mine, Nat
arrived, good to catch up with him all too quickly, taxi
to train etc.
- Slept on the train a fair bit, with a little
hacking, it seems that generating a load of new /
custom relocations may be a little tricky: it'd be far
easier & more efficient to get the data into a form
that glibc can understand easily (and can be .rodata) -
unfortunately of course it seems acutely difficult to
get any change no matter how compelling into glibc to
improve linking: indeed even discussing such a
change is ~impossible. The powers that be simply refuse
to communicate meaningfully.
- Kissed sleeping babes generally, bed exhausted.
- Up lateish, caught the end of breakfast; taxi
with Miguel to the conference, met a load of fun people.
Simon Phipps' talk - quite fun, more hacking in the back.
- Fun packed day, lots of people to meet &
catch up with, fun - dinner with Miguel, Pippin & misc.
speakers in the evening - learned a lot including the Go
rules. On to a pub with a bevy of Sun & Java people -
met various Mark's, Miguel arrived to discuss Mono / Java.
- Up earlyish, breakfast with the family - off to
the Funky Fun House, with Naomi Brighty. A suitably
fun time had by all little people; taxi to station, train.
- Caught the Eurostar eventually - encounter with
a bevy of open-hand(?) hackers, pleased to meet Tom Tromey
(of old-time Gnome fame, now gcj), and Andrew Haley - spent
the trip discussing gcc, java, OO.o etc. Taxi to the hotel
with Jerve.
- Looked up
Polymorphism to check my recollection of the definition -
mercifully overloading is (ad-hoc) polymorphism.
- Some binutils hackery, met up with Miguel, more
binutils bits: got the entries sorted (very stupid algorithm),
got the data read from the section: looking rather pleasant.
- Up early, breakfast, talked to Florian much of
the day , chewed over various writer topics,
caught up generally.
- Back to Hahn just in time for the flight - more
binutils hackerage on the plane - got the symbol information
I need in the right place: great. Now to build a tree of the
symbol information to let the inheritance graph inform the
copy order - so we don't copy things before we have them
in place.
- Drove home, arrived eventually - dinner with
beautiful wife, and small bouncing beans, (also Janine &
kids), conference call, J. out to bible study, more work:
tidied loose ends, discovered text messaging etc. Marveled
again at the quality & volume of work from Petr. Caught
up on bug mail variously & filed a few more myself.
- I like Mike's idea for OO.o Janators projects,
some tasks might be adding visibility markup to libraries,
writing unit tests, removing obsolete platform conditionals,
warning reduction, seek/destroy header conditionals hurting
the PCH work, etc. etc. Worked late.
- Up early, packed babes off to pre-school, poked
mail, planned day etc. Drove to Stanstead, checked in, hacked
for a while - slightly nervous wrt. Ryan-air's idea of a
"Final boarding call" 35 minutes before they finished
boarding: if there was laptop power near the gate or on
the plane: people might get there earlier.
- Hacked at binutils - much simpler code than gcc,
almost fun. Flight, more hacking - eventually despaired of
finding the right point where the various structures are
populated with the information I need, and set out doing the
reloc parsing more intensively myself.
- Bus ended up at the Hauptbahnhof - Florian picked
me up, off to a pleasant restaurant with Mahren too, nice to
get to know her a little. Back to the hotel lateish - caught
up with the world via. E-mail, GWIM, more mail, poked people,
re-booked flights tomorrow, bed too late.
- Up early, breakfast with Robert. Poked the
mail, quick call with Fridrich. Poked at some objdump
output to see if we can catch a host of 64bit polymorphic
virtual overrider issues automatically.
- Turned some paperwork around festering in my
inbox; back to binutils. Poked at crtstuff - perhaps this
is a good place to do the work.
- Sandy popped around, last minute call, went
on into the evening.
- Up - poked at mail, back to gcc: fun, it's
unfortunate that it's not possible to use gcc's own RTL
generation from parsing (etc.) to auto-generate more
complex types & code fragments internally, luckily
-fdump-tree-all output is quite useful.
- Plugged away - managed to get most of it in
place before hitting some other grisly internal
type-checking issue with hand-generated trees. Lydia came
over for lunch.
- Some great news Kohei will join Novell from
March 1st, another of our ongoing hires to work on OpenXML
support built into OO.o / calc. Kohei is famous for his
OO.o solver
hacking along with some other calc bits. Really looking
forward to having him on the team full-time.
- More gcc bits, interview. Managed to generate
a first cut of the data to inform the vtable copying
process, hopefully binutils can be persuaded to swallow it.
- Out to the
Collabora office to see Rob & Dan Williams, meet
Dafydd & Gill(?). Poked at the profusion of one-laptops,
and Nokia devices, admired various features. Out to the Free
Press for a meal, joined by Matthew Garret, good stuff.
Back, Robert staying overnight, bed late.
- Up early, dealt with babes, off to NCC - Tony
speaking on faith. Back for lunch with Jeff, Donna, Isabelle
& Jake.
- Watched DVD with the babes in the attic, dinner,
bath, bed early, couldn't sleep.
- Up very late, wonderful lie-in from the wife;
lunch, poked mail - filed calc bug, wrote status report.
Read up on binutils linker scripts, discovered collect2;
'info gccint' is rather interesting.
- Up early, kept babes quiet while Mum slept for
some time; up, out of the house for some cycling practice on
the gallops (working late tonight). Back, early lunch, to
work.
- Filed various bugs, updated packages etc. Worked
away at gcc - got some nice vtable data spewing out on the
console, onwards to binutils to see how to generate some
suitable global constructors.
- Call in the evening & note writing.
- Up early, prodded mail gingerly not much; good.
Onto the gcc work, very strange things lurk here; diff -u on
the generated assembler quite fun. Discovered that at least
one cause of problems is some random element to symbol names
(from anonymous namespaces ?):
_ZN83_GLOBAL__N__opt_OpenOffice_OOE680_m6_sw_source_ui_uno_unoatxt.cxx_00000000_E6D2AB5A12rtl_InstanceIN4cppu10class_dataENS1_14ImplClassData3IN3com3sun4star4text18XAutoTextContainerENS6_4lang12XServiceInfoENS6_9container12XIndexAccessENS1_15WeakImplHelper3IS8_SA_SC_EEEEN3osl5GuardINSG_5MutexEEENSG_14GetGlobalMutexEiiE6createESF_SK_
(eat that c++filt). Also discovered that an in-directory
configure creates it's own separate build-dir: host-...
critical for fast incremental builds; nice.
- Interview, lunch, gcc defying me. Wasted the best
part of an hour trying to track down a hideous hard-kernel
lockup when beating the system with valgrind / OO.o. It defies
the nmi_watchdog, and (worse) appears isolated to when the ATi
binary driver is in use: go binary drivers !
- Dinner in 15mins, another late call, story reading
for children, bed early.
- Up early, Valentines day - cards & chocolates
for the darling wife. In deference to the Dutchess of Cornwall
it appears Hannah wants a Sister-ectomy, "I want to be an
only-child", "Don't you want Myriam ?", "I don't mind her
it's Naomi!" clearly the strain of being treated equally
as one three is harder with a small age gap. Luckily only a
transient aberration, they love each other really.
- Poked mail, updated to HEAD ooo-build & started
a build, cleaned up some lurking unused patches. Mostly thanks to
getting Noel (& Jian & Zhang)'s VBA work up-stream, and
also due to Rene's up-streaming efforts, we're finally showing
some decisive progress in shrinking ooo-build:
at least by some (vague) measure of
individual patch count. It's worth pointing out that of these
outstanding patches 346 are filed up-stream, the remaning 194
having various issues / being too controversial for up-stream
(eg. defaulting to using the system file dialog).
- Back to mail, poked into a bug or two and an
autobuild build error. Spamed the interface-announce list as
requested wrt. 'intern'.
- Lunch with heart shaped jam tarts, neat. Paperwork,
bother, chat with Miguel - unwound strange pyuno related
installation breakage in HEAD: UNO/python stuff appears to be a
perennial pain wrt. obscure component registration failures.
Turned out to be some unrelated environment change.
- Call with Patrick & Nat, managed to over-run my
pre-announced early stop by 30mins, frantic bathing of children,
reading of stories; Mary & Lydia arrived for dinner, Cell
group, bed late tired.
- Up early, off to NOD to fix their networking - again
someone fiddling with the wires to the hub: clearly we need a
lock on the cabinet door to keep enterprising, homeless teens
out of the cabling.
- To work - poked mail, a day of paperwork looms.
Conf call at some length. Booked travel. Finally set off an 'sw'
build with the hacked gcc to cripple vtables to not have empty
inherited slots.
- Kelli's staff late.
- Up early, call with Nat, poked Katy, rummaged In box.
Into town to buy a new ledger, can't have my life strewn across
the floor on pieces of paper. Managed to stock up on the obligatory
gifts for the other half (absent minded husbands beware: Feb 14th
looms).
- Poked at yast bits that Jiri sorted out, looking good,
poked at an ORBit2 issue. Great to see NN blogging about his (great)
calc work, and look forward to seeing our GETPIVOTDATA patch out of
ooo-build. Annoyed by my electric fan-heater; it appears the
hysteresis on the thermostat is just larger than my built-in one,
leading to removal of clothing just before the thing turns off &
vv. Upgraded to latest SLED10 devel packages.
- Waded through some rather unpleasantly complicated code
in SfxMedium land, and back to reading gcc vtable construction
logic, talked to Michael Matz. Fixed the salstrintern regression
tests.
- Up early, tried to encourage children to remain quiet
while Mum slept, with limited success. Sustained quietness is not
something that comes easily to little people it seems.
- Off to NCC - Janice spoke well, using a number of
Brick Testament slides: a rather esoteric interpretation of
scripture for sure.
- Back for lunch, babes slept, then cream-teas with
Kate & Roger, then the Brighty clan until late - tea with
them, bed earlyish.
- Up in the night with mail, and up again rather late.
Breakfast, put babes to bed.
- Side-tracked by yast2 bits, brushed up my flaky
shell a little. Lunch, interesting conf call.
- Some wonderful news: Today, Sun
released their proprietary VBA macro migration plugin as Free
Software, and committed to working with Novell to integrate it with
our code, to improve interoperability for all OO.o users. This is really
good news, and (I think) part of Sun's increasing focus on OO.o
vs. StarOffice. I'm looking forward to seeing the fruit of the
collaboration, and getting this integrated up-stream quickly. Also VBA
interop is an area where new hackers can quickly get involved, writing
the mapping code is not only immediately rewarding, but is often easy
particularly porting & integrating Sun's existing helper API. In
conclusion: this is excellent news for users, and I'm extremly happy
to welcome Sun to the OpenOffice.org VBA interop. project with Andreas
Bregas set to join as co-lead. It's just great to work together on this.
Thanks too to Juergen for all his work making this happen. Oh, and for
those that care: as part of the up-streaming process, this piece will
become fully pluggable.
- Mike Fabian fixed my fontconfig crasher, tested it
works beautifully; wonderful.
- Got poked by two people internally asking about Sun's
"announcement of OpenXML support" - it -appears- that they're mistaken
and talking about
the MS Office ODF Plugin - which of course is totally
different. To re-emphasise our position: while I'm thrilled that Sun's
MS Office plugin is re-using the OO.o code-base: the best way
to get perfect interoperability is to use OO.o on both Win32 and Linux
and exchange documents in ODF format. ie. if it's at all possible: abandon
that MS beast, and move to a Free software alternative; don't just
plug some Free-software into it to tame it slightly.
- The wonderful wife produced a pudding this evening: nice,
had to skip 1/2 of the 1st course to write mail.
- Up early, no snow, few minutes later: lots of
settling snow: sent the wife out in it; hmm. Velux covered in
it so difficult to see out.
- Pleased to see Simon's blog on Sun's up-coming
Odf Converter. As an avid admirer of Brooks, I'm still
convinced that if there is a silver bullet it's code re-use;
were I to design such a plugin, this is how I would do it too:
the more we re-use OO.o around the place, the more we are going
to have to re-factor and improve the code, and the more
stake-holders get interested in contributing to that. Rock on
OpenOffice.org.
- Poked mail, poked at an a11y issue, admin: wrote
some copy, tried to schedule a meeting. Committed a psprint
fix to remove %%Title generation.
- Built a snow-man after lunch with the babes in
the road, some abhortive (and somewhat one sided) snow-ball
fights.
- Massaged salstrintern into some better shape,
started adding other modules to it to add the small tweaks
around the place that will substantially reduce our string
duplication memory overhead. Cleaned up all Stefan's concerns,
measured the memory win: 850K (pmap dirty). Set off a test
build. Poked at unit tests in configrefactor01.
- Dinner, with Janine & kids, put babes to bed
read stories, washed up, assembled (still unassembled) birthday
present (stroller thing) for N. (Mouser), removed glue from
hands, poked at laptop.
- Fixed silly gvfs bug - silly infinite recursion
instead of chain to another method, further validating my
contention that the Stream API method 'readSomeBytes' is not
that useful an API vs. 'readBytes'. Played with test build,
marked salstrintern 'Ready for QA'.
- Up early, little mail. Submitted yast2-gtk with
various fixes to autobuild, poked at my strange fontconfig
crasher.
- Got a lovely patch from Stefan to fix / improve
some configrefactor01 issues, resynched to m202, committed it.
Tried to get the DSL router configured correctly, eventually
managed to reset it, including all the setup information
- urgh. Eventually got it sorted out again, pointed Mike Fabian
at the relevant fontconfig issue.
- Rescynched salstrintern to m202, backed out the more
controversial in-line string bits etc. fixed ~all of Stefans
issues, and poked him again.
- Poked mhu wrt. .rdb file savings, and started digging
into gcc's guts for the vtable copy construction bits.
- Out to cell group, with Robin at Shelly & Pete's,
small but fun, did a random Thessalonians study by mistake (took
the wrong book). Back late, very cold out.
- An interesting UE paper Federico linked to:
cgis.cs.umd.edu. Word Processing at the top for the
"Workplace Study causes of frustration". Table 5's average
minutes lost per frustrating experience rather amusing,
and the paragraph below Table 1 In only one [ of 149 ]
frustrating experience did the user consult a manual, and in
only two experiences did the user consult on-line help.
Is it possible the concept of blocking lots of scattered UI
improvements on the availability of UE feedback, and complete,
translated on-line-help is an exacerbating factor in user
frustration ? To paraphrase mhu: Are we retarding optimising
the 98% in order to focus on the 2% ? Perhaps things changed
since 2004 though.
- Dug at mail. Encouraging chat with Michael.
Unwound various package related problems with banshee. Looking
at bying a DLP Projector for personal use to entertain people
with movies at home: needs to be at least 1024x768, so far
the Dell branded beasty seems a good price, am I missing
something ?
- Got two almost duplicate mails from people pointing
out how much OO.o release notes could be improved - hopefully
they'll put in the 30mins / month necessary to fix that.
- Spent all of the hacking morning banging away on STL,
reduced to 'trying' things: depressing, the combination of incredibly
verbose & unhelpful gcc error messages, unreadable and unhelpful
implementation, and 'meta'
documentation that prefers to dispense gems such as
"Antisymmetry is a theorem, not an axiom: it follows
from irreflexivity and transitivity." rather than concrete
samples. Having said that I was wrong wrt. binary sort - Thorsten
put me right & helped unwind some of the more evil problems,
thanks lad.
- Some particularly good news today though, after
a rather protracted hiring process (sorry lad) we got Fridrich
on board with Novell, starting 1st March, and we're looking
forward to having him bring his impressive WordPerfect filter
experience to bear on importing OpenXML. Also, hopefully it's a
sign that hacking on OO.o as a volunteer can pay off (eventually).
- Team meeting after lunch, nice to have JP to run it.
Dinner in the evening. Hacked at yast2 in some spare cycles - got
YGCheckBoxFrame implemented, so it builds again.
- Poked at a postscript generation 'bug', some more
travel pokeage.
- Somewhat appalled by the OO.o 2.1 comments in
LXF 90 by Cynux: ... Firefox going places. On the other hand,
OOo is stuck in neutral - it took a year to move from 2.0 to 2.1
and the headline new feature is multimonitor support for
Impress. Clearly a tragic failure by marketing to actually
communicate the fact that this is the 1st of regular
releases to use a new versioning scheme to show the fact
that we include new features as well as bug fixes. ie. people
shouldn't expect some huge bunch of features in 2.1 ? why not ?
because we produce features at a fairly constant rate: so if
you want a huge bunch of them piled into a single release, that
means we had to dam the feature stream for years, holding stuff
back from the end-user. On the other hand, it's a shame our 2.1
announcement / text on the front-page didn't correct this
(all too predictable) misunderstanding, but instead says:
"OpenOffice.org 2.1 is recommended for all users,
as it represents a significant improvement over all
previous versions...." etc. In fact, IMHO the regular OO.o
releases are an encouraging sign that OO.o is out of neutral
and pulling into the fast lane.
- Also - if marketing types are reading: SoftMaker
have
this horribly selected & inaccurate, but compelling
web-page. Jody is working on #3, the last 4 need to wait for
chart2 etc., but the presentation is nice. Surely we have some
talented people that can create some similar nice samples
and create the relevant screenshots to compare MS Office, OO.o,
Softmaker, and any other of our competitors ?
- Updated to the latest SLED10 test packages. Committed
more code to configmgr, hopefully saving another 50k+ or so of
redundant STL RBTree nodes, now using a binary search of a flat
vector (something that apparently STL can't help with): strangely
memprof predicted a larger win; odd.
- Call from Peter, built HEAD YaST2 idly, fixed the
control-center-gnome package to build in a more 'yast-like'
fashion.
- Up earlyish, J. slept, fed babes, to NCC - Thea
speaking, back for lunch with Armenio, Rose & Victoria -
good to get to know them a little better.
- Sue & Clive dropped in for scones on the way
home, lovely to see them too. Bed early.
- Up late, breakfast, a little
mail pokeage while babes slept, fine lunch, down to the beach,
wandered along by Sizewell powerstations.
- Back, slugged pleasantly, drove home late, bed.
- Up rather early, off to the Eurostar, still
digesting last night's meal; hmm. Train, tube, train -
hacked at configmgr a little - yet, yet again frustrated
by STL: inefficiency and lack of usability combined
in one big package.
- Back to Bruce & Anne's - lovely to see the
little girls again, lots of climbing up Daddy etc. Back to
work eventually, dug at the mail. Caught up with JP, read
some nice blurb from Jurgen, poked Noel, good stuff. Then
a good chunk of faffing about.
- Up, breakfast with Gilles, Lars & Roman. Off
to do talk - Jo did a cool smart-card enabled desktop demo;
wandered the show floor, caught up with Dave Neary, and Roman,
had my (first-ever?) manicure; hmm.
- Back to the hotel to hack, apparently some switch
died, so nothing is working properly. Apparently LXF column is
due today too. Hacked on that, lunch with Damien.
- Back to work, then to the booth, caught up with Lars
& Jo De-Baar. Guy arrived, and headed to the Novell office
for some hackerage while it got dark, great to catch up with Guy.
- Out for an extensive French meal at one of Guy's
old stamping grounds with Holger & Michael too; lots of fine
things to eat & drink.
In case it's not painfully obvious: the reflections reflected here are my
own; mine, all mine ! and don't reflect the views of Novell, The
Lithuanian Gov't or Arnold Schwarzenegger. It's also important to
realise that I'm not in on the Swedish Conspiracy.
Occasionally people ask for formal photos for conferences,
bio.
or fun.
Michael Meeks (michael.meeks@novell.com)
