i should really go to bed.

to wit: GType gtk_entry_get_type (void) G_GNUC_CONST;

glib's own documentation has this to say on the matter:

Likewise, a function that calls a non-const function usually must not be const.

"usually must not" is a little passive-aggressive for api docs, but anyway these _get_type() functions almost all call g_type_register_static_simple, which is not const.

anyway, the problem is if you have code like this:

GTK_TYPE_ENTRY; g_print ("%s\n", g_type_name (g_type_from_name ("GtkEntry")));

it may or may not work, depending on your optimization level. if you ignore the return value of a G_GNUC_CONST function, gcc -O2 will happily optimize it out, as i discovered long ago.

the fix workaround is to pass the return value from the const call into a function that is not declared const; g_type_qname() fits this bill rather nicely.

well, i'm sure they have a good reason.

it's still sort of weird working for a proprietary company. just about nothing i worked on at novell ever got released, or looked at by anyone other than myself, so i didn't think it would be a big change. but at an employee meeting the other day they were recognizing people who'd registered software patents. the juxtaposition between the cheering there and the reaction most of my normal friends have to software patents is quite striking.

basically, it lets me register an object that gets called when elements with a specific namespace are encountered in a document, and i can create an object that is effective aggregated with that element. since i use namespaces, this is exactly what i have been looking for this whole time. and it looks like it'd be pretty easy to add support for other libraries, if you are interested in some incredibly erotic widgets.

unfortunately, though, this means throwing out just about every line of code i've written since october, and more or less starting anew. the cost of progress, i guess. and anyway, did i really want to maintain my own dom implementation?

anyway, in a night or two i got it creating widgets, and setting their properties (including container child properties) working correctly. i even have an nsIDOMNavigator implementation written in javascript. pretty awesome! anyway, i think the difficult legwork in getting scripts working is done, but there's a little more work there that will have to wait for tomorrow.

i am not quite yet ready to svn rm my old implementation; but that day is fast approaching...

of course, that was two mergers ago, and compusa isn't even around anymore.

anyway, i just closed my bank of america checking account. hooray!

what's even more ridiculous is that distcc is already open-source. what the hell, guys? multicast dns-enabled (x)inetd probably is, too. good going.

their faq is full of awful english, and even worse jokes; i will not dignify it with a link, but two elements of FUD do merit addressing:

you're changing compiler versions often and still want to speed up your compilation (see the ICECC_VERSION support)

i... i don't know why you are changing compilers often. i am not going to ask; i really don't want to know. anyway, if your company hasn't figured out how to distribute a toolchain to its developers, dag has awesome distcc compiler packages for red hat and fedora. i used to use these when i had a mini cluster in brookline, and it was the best. cross compilers were no problem, either; i also distributed x86 builds to my ppc macs.

most important: you're sitting in a office with several co-workers that do not like if you overload their workstations when they play doom (distcc doesn't have a scheduler)

it's 2008, guys. if your workstation can't handle DooM on one of its cores while the other three are compiling compilers, well, i don't even know. maybe if you had a stable kernel abi that didn't require recompilation of drivers for SMP and UP machines, you wouldn't have to spend so much time compiling? just throwing that out there.

also, it cannot go without saying that today was the first boston radiohead show of the millennium that i missed. and they played kid a and a wolf at the door! i can take some solace in the fact that i wouldn't be going to the smelly old tweeter center anyway, even if i wasn't in california.

i will also take this opportunity to mention that tickets for radiohead's show IN TOKYO ON MY BIRTHDAY (WHICH IS ALSO THOM'S BIRTHDAY) go on sale august 23. you know, if you happen to be shopping on the internet that day.

that was all well and good, until i got it working, and tried to use XMLSerializer to print out my dom tree. that did not work. you see, there is a pesky little function call nsContentUtils::CanCallerAccess(), which contains this:

nsCOMPtr<nsINode> node = do_QueryInterface(aNode); NS_ENSURE_TRUE(node, PR_FALSE);

unfortunately, while my objects support nsIDOMNode, they do not support nsINode; its header is not in the public sdk directory, and besides:

class nsINode : public nsPIDOMEventTarget {

that P doesn't lead me to believe i should be implementing it.

ok, that's fine. i can create a dummy document with DOMParser, and import my document into it, and then dump that.

of course, nsDocument::ImportNode() has this:

nsCOMPtr<nsINode> imported = do_QueryInterface(aImportedNode); NS_ENSURE_TRUE(imported, NS_ERROR_FAILURE);

i'm sure i'll think of something in my sleep.

maybe it's also regenerative?

this is quite different from what i was doing before; i'm now using a "real" Mozilla XMLHttpRequest object to fetch a gom XML file. then, i use my dom implementation to import the XHR's responseXML document element into a gom document.

it is basically magic.

i've had the guy from lifestyles of the rich and famous in my head all night, describing gom as "home to the world's most gorgeous interfaces." i don't know, it's just there.

it's not ready to commit, but after days of deliberation, i think i'm narrowing in on a nice xpcom/gobject bridge, with this clever hack at its heart:

gpointer gom_unknown_query_interface (gpointer unknown, GType requested_interface, GError **error) { if (GOM_UNKNOWN_GET_INTERFACE (unknown)->query_interface) { return GOM_UNKNOWN_GET_INTERFACE (unknown)->query_interface (unknown, requested_interface, error); } else if (g_type_is_a (G_OBJECT_TYPE (unknown), requested_interface)) { return g_object_ref (unknown); } g_set_error (error, GOM_DOM_EXCEPTION_ERROR, GOM_NO_INTERFACE_ERR, "Object does not support %s", g_type_name (requested_interface)); return NULL; }

the only class that will actually implement GomUnknown's query_interface is my base class for GObjects that wrap an xpcom object, which is:

static gpointer gom_wrapped_query_interface (gpointer unknown, GType requested_interface, GError **error) { GomWrappedPrivate *priv = PRIV (unknown); if (g_type_is_a (G_OBJECT_TYPE (unknown), requested_interface)) { return g_object_ref (unknown); } return gom_wrap_xpcom (priv->wrappedObject, requested_interface, error); }

gom_wrap_xpcom() walks my list of IID <-> GType mappings, and will return a (possibly cached) GObject that implements requested_interface by simply calling each function's xpcom equivalent.

maybe i should have just used python.

i have been very hungry the past couple of days, though.

that's what i figured.

still sore. so sore.

oh my god, i almost forgot. before the race yesterday, the MC was doing some thing about how some troops ran "the san francisco marathon" from afganistan, or something, whatever. then he said we hope they'll come home soon, wherever they are: afganistan, iraq, or iran.

ahem.

• january 2000: i move to boston
• december 2000: manny ramirez signs with the red sox
• june 2008: i move to san francisco
• july 2008: manny ramirez is traded to the dodgers

LA isn't quite SF, but still.

they have some sort of real-time text messaging thing, so if anyone wants to follow along in my brazen disappointment, let me know and i'll add you to the thing. for those of you who enjoy sleeping late on sunday, note that it starts around 5:30 am pacific.

other events that have recently transpired have been the 11th anniversary of the last time i drove a car, and the receipt of a letter from my health insurance company with the following amusing excerpts:

Some medical conditions, either alone or in compination with the cost of medication, present uncertain medical underwriting risks. In view of these risks, we have enrolled you in our CORE 5000 PLAN Level 1 Plus 50 Plan. This is our CORE 5000 PLAN Level 1 Plan with an additional 50% premium. The basis for our decision is listed below.

Provider: Health Statement
Medical History: Elevated weight

...

To be considered for a reduction in the additional premium, you must 1. maintain your weight below 215 lbs for a mimimum of 6 months from the effective date of this coverage, and 2. meet the Medical Underwriting Guidelines in effect at the time of application.

i think i'm up to the challenge.

MARGARET WARNER: Bottom line, Mr. Prime Minister, what could the U.S. expect to see, let's say in six months, as a benchmark, as evidence that your government's approach to fighting terrorism is more successful than President Musharraf's was?

YOUSUF RAZA GILANI: I really don't know. But at the same time, I think, when there is any actions that has the support and the backing of a political government, it is always successful.

MARGARET WARNER: But have you set any benchmarks for yourself?

YOUSUF RAZA GILANI: Certainly that we are convinced ourselves that it's working.

...when there is any actions that has the support and the backing of a political government, it is always successful.

well, i'm convinced, too.

it was a bit painful at times, yes, but not nearly as frustrating as trying to get the python bits working (despite that being three-year-ago-me's fault). it's even now using the standard gecko sdk which should make it easier for others to build, and me to support.

after i had written gom_uri_get_contents(), and everything was working great locally, i wondered: do i now support remote files?

well, as long as it doesn't require an auth dialog or anything.

i guess next up is xpconnect, which gives me XHR for free?

i had hoped that libcurl would suffice, but it simply doesn't export this functionality (although it must implement it internally). sad. i couldn't find anything in the new gio apis either, which was disappointing as well.

so, i went looking. the first one i found was uri. unfortunately it seems to have been a victim of the warn on terror, so i moved on.

next up was uriparser. it has a promising name, and does what i want, but the API is sort of dreadful, and it still had crashable bugs as of 2008. the final kicker was that although it supports ACII and Unicode (UCS-4, i think?), it doesn't seem to want to do UTF-8 out-of-the-box, which is a non-starter these days.

(unicode things still confuse me terribly, so maybe i am wrong here)

i next considered trying to use bits from mozilla, but the necko docs yell at me, and then throw me at this. i was hoping for a little more hand holding.

what i really want is GnomeVFSURI. of course, gnome-vfs has been deprecated for years now, but still there is nothing equivalent that i could find. i didn't really want to depend on gnome-vfs, so when ports wanted to install its own version of perl i lost it and started trying to get gnome-vfs-uri.c to build as a library on its own. this was going nowhere, fast.

then i remembered that i asked someone at work, and got the usual "Oh just use Python, it's so great." i wasn't keen on adding a third type and memory management system, but the api was great, and what i saw of embedding python from hula wasn't so bad.

well, of course i was up all night, trying to find out what @parameter meant, and why i could see properties attributes when i ran the thing in python that i couldn't see when i ran it from C. why would they make so much more difficult to access attributes from C than it used to be? why am i such a terrible programmer? will i actually fall asleep if i go to bed now at 9 am?

eventually, i noticed the following:

(gdb) p PyString_AsString(PyObject_Repr(mod)) $24 = 0x171e5b4 "<module 'urlparse' from '/Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/urlparse.pyc'>"

this didn't look right. in python:

>>> repr (urlparse) "<module 'urlparse' from '/System/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/urlparse.pyc'>"

of course; this API is different in 2.4 and 2.5; i just rued again. even after completely obliterating the 2.4 version of python (don't even ask why i had some third-party version installed; i certainly have no clue), if i have to do some compat stuff for the 2.4 version, it's just not worth it. i checked it in for now since it works, but i still need to find a permanent solution.

if it wasn't so completely beautiful outside, i would go get an xbox and gta4, but i think i'll do some grocery shopping instead.

it's a good thing i didn't give them my SSN.

anyway, i put up a new release before i go and implement XHR.

the other day i applied for health insurance. there were pages of online forms to fill out, and they ask you to verify that the information is correct, but i wasn't paying attention and didn't notice that i accidentally misypted my weight - reporting myself as 100 lbs heavier than i actually am.

i tried to think of a better way of making people verify their forms, and i came up with the following: go through the list of fields randomly, and present the user with their answer for that field. below this, offer them a choice of 4 possible options as to what that field is. it's pretty easy - they can recognize that their street address is their street address - and i think it would be easier for them to spot typos this way.

i'm beginning to wonder whether GObject/C was the best choice here...

since it's been a while, i'll refresh your memory:

What is Gom?

Gom allows you to write GTK applications using JavaScript, an HTML-like widget layout syntax, the standard W3C DOM interfaces, and jQuery.

Gom is on the Web! http://ilovegom.googlecode.com/

next up has to be XMLHttpRequest.

i've been working at vmware for about a month now, and i don't think i've gone to the same place in the city for lunch twice yet. that is certainly something.

and if those aren't enough layers of abstraction from X or Quartz, you can always throw your favorite javascript library on top of it.

all this, and you don't need to deal with building on different platforms and distributions!

at least that's the idea.

last night i finally checked in those changes i mentioned. i then went late into the night trying to track down numerous refcounting and garbage collection. getting close.

the next three weeks of training are fairly inconsequential and will hopefully be without incident. there isn't much more i can do but wait it out, and we'll see if i trained hard enough. yesterday, i would have said "yes" but today i'm not so sure. of one thing, however, i am certain: i will not be riding my bike home from work the week before the marathon. i could feel the effects of the hills thursday and friday; in fact these hills are for the birds.

for the first time in almost seven weeks, i am typing this at my desk. the bookshelf is now stocked with books as well; i have an office! four boxes remain unopened (roomba, scooba, rock band, guitar hero II), while four more boxes of wires and doodads remain. it could be said that i am done unpacking; the rest is "merely" putting things away and cleaning up.

tomorrow i will put up my art and there will be no mistaking: this is my apartment.

now it's time to go wallow in bed for a few hours, rueing the day i ate five packets of caffeinated gu.

anyway today i had all the foods i love the most, in preparation for my insane run tomorrow. breakfast was two of the following: eggs, toast, bananas, and glasses of milk. dinner was a ton of pasta, a banana, and two more glasses of milk. i don't know what it is about these foods, but they make me feel happy from the inside out.

of the boxes that i will actually be unpacking, five seven remain unopened, and there's another 4 or 5 full of wires and stuff that haven't been unpacked. just about all the furniture is put together (somehow i managed to get my desk together by myself today), and it's starting to feel a little like home. i need to get some sort of bike rack, because storing my bike in the bathroom isn't going to cut it if i ever have guests.

i got a crazy messenger bag today. it is wicked orange.

this is part of my new initiative, to have at most one computer running. this is a new direction for me; who knows where it will lead.

(of course, there will still be a tivo, slingbox, and two (down from five? (!!)) airport expresses always running...)

unpacking is going well. the boxes that will never be unpacked are starting to trickle into the living room closet, their final destination. i am starting to accumulate a huge mound of "wrapping" paper in the breakfast nook office. i don't know what to do with it.

i cycled to the office today, and it wasn't nearly as scary as i feared. however, the hills coming home are moderately fierce. it has occurred to me that perhaps i should wait until after the marathon is over before i make this a regular occurrence.

if the tie doesn't fit, you must acquit?

i read a bunch of books on my kindle while i was in europe. here is a brief list:

1. Do Androids Dream of Electric Sheep? - Philip K. Dick
2. Brave New World - Aldous Huxley
3. Nonzero: The Logic of Human Destiny - Robert Wright
4. Pride and Prejudice - Jane Austen
5. A Tale of Two Cities - Charles Dickens
6. Alice in Wonderland - Lewis Carroll
7. Through the Looking Glass - Lewis Carroll
8. The Memoirs of Barry Lyndon, Esq. - William Makepeace Thackeray

i haven't quite finished the last one there, because i have been reading the sf chronicle (i'll give you a minute to groan. go ahead, i'll wait. the thing is, the times just doesn't have enough articles on public transit or earthquakes). anyway, for one reason or another, i've managed to highlight the following articles:

"We must embark on a national mission to eliminate our dependence on foreign oil and reduce greenhouse gases through the development of alternative energy sources."

---

Fisher acknowledges that he is not well known, has never built a skyscraper before and hasn't practiced architecture regularly in decades.

---

A study released earlier this year by a group found that increased driving as a result of sprawling development in the next few decades will overwhelm any gains from increased vehicle fuel economy and use of low-carbon fuels like biodiesel.

---

For McCain and Obama, the economic message should be that "we can't make (Americans') house value go up and gas prices go down, but we can improve education, work on alternative energy and invest in infrastructure. And probably if they weren't in the middle of the campaign, they would be a lot closer on doing that."

---

A security researcher revealed a flaw Tuesday that makes it possible for a hacker to take control of the Internet.

anyway, time to do dishes.

so whatever, i skipped it. today was the last day at the track before the race, and i'd say it went well (12x800m):

1. 3:05
2. 3:03
3. 2:58
4. 2:55*
5. 3:02
6. 3:02
7. 3:03
8. 3:08
9. 3:05
10. 3:08
11. 3:02
12. 2:58

(* = new PR)

and then after work i almost missed the late shuttle back to the city and had to run to catch it and that wasn't fun.

i'm finally feeling like i'm turning a corner in the unpacking. i think there may have to be a craigslist firesale after all, if phik doesn't claim my shuttle first.

i am so tired and want to sleep so badly but i am going to be an adult and go wash that one dirty knife i used to butter my toast.

five miles of my run this morning was spent along the beach. there were hundreds of washed-up jellies. it was gross.

there was also one huge seal (or sealion?). it was even more gross.

as an update to the comcast debacle, they've already sent me a bill. it has a credit for $20, but not the full $100 that i was under the impression i was getting. i guess i get to call back monday.

it's also night one of living without a dishwasher. i am super tired, and really want to go to bed, but i better wash things from dinner...

and that's pretty embarrassing. so i tried to fix it, now that i have unpacking to put off.

patch one wasn't too difficult to track down once i started reading some of the code where i was getting the failed assertion. it was the classic passing an int instead of a jsval (even though my getter func doesn't look at that field; it just returns the "tiny" id):

Index: src/libgom/gomjsobject.c =================================================================== --- src/libgom/gomjsobject.c (revision 57) +++ src/libgom/gomjsobject.c (working copy) @@ -602,7 +613,7 @@ JS_DefinePropertyWithTinyId (cx, ctor, &enums->values[i].value_name[4], enums->values[i].value, - enums->values[i].value, + INT_TO_JSVAL(enums->values[i].value), gom_js_object_get_enum, NULL, JSPROP_READONLY | JSPROP_PERMANENT); }

for the other, i pulled out all the stops, from hardware watchpoints to malloc debuggers, to that awesome gdb trick, to not installing a linux vm to use valgrind and refdbg on. i went back to the tried-and-true method of littering code with printfs, and narrowed it down to someting with the Event.target field. eventually i realized that while i had written a great finalizer for my base class, none of the subclasses chained up the finalize call, which led me to return dirty, used, and freed objects later. my classes have a lot of this now:

Index: src/libgom/gomjsevent.c =================================================================== --- src/libgom/gomjsevent.c (revision 57) +++ src/libgom/gomjsevent.c (working copy) @@ -34,6 +34,12 @@ #include <glib.h> +static void +gom_js_event_finalize (JSContext *cx, JSObject *obj) +{ + GomJSObjectClass.finalize (cx, obj); +} + JSClass GomJSEventClass = { "Event", JSCLASS_NEW_ENUMERATE, @@ -44,7 +50,7 @@ (JSEnumerateOp)gom_js_object_enumerate, JS_ResolveStub, JS_ConvertStub, - JS_FinalizeStub + gom_js_event_finalize }; static JSPropertySpec gom_js_event_props[] = { { NULL } };

tomorrow, instead of unpacking my dishes and glasses so that i could make dinner, i will remove all the cruft from my tree, and finally make that third release i've been wanting to do for three months.

also, if i didn't fix this bug i don't think my mind would let me sleep tonight, so i am doubly excited. wooooo WOOOOOOOOOOOO!

as a bonus for taking the day off for moving stuffs, peach came to the city and we got to hang out for a little. neither of us could remember seeing each other since i was here before i started at novell, but it couldn't have been three years. could it?

i barely got 4 hours of sleep last night, in two shifts, but i managed to have a funny dream. i was trying to get away from having damaged a teacher's car with a tree (long story), and i remembered i could just go to my room upstairs (as in, i was in my mom's house). the bed up there had blankets, and was covered in clean laundry that needed to be folded. i will leave as an excercise for the reader to find all 8 levels of wish-fulfillment in that one.

Deposit paper checks - You can also deposit paper checks one of two ways:

1. Deposit the check into your linked checking account then move it to your Electric Orange...
2. Mail in checks

that seems lame. the credit union checking account doesn't seem to have a direct-deposit activity requirement like one of BoA's checking accounts (although i suppose i could have it deposit only $1). BoA's other checking account doesn't have a direct deposit requirement, and has a lower minimum balance than the credit union, but at least the latter pays interest and doesn't give BoA any more of my money.

so, at least for a little while, i'll keep the credit union account. anyway, i suppose it can't hurt to have a little emergency fund tucked away.

in a little under 12 hours, the movers should be here filling my new apartment with way too much stuff i don't want anymore, and a few things i do want. i am looking forward to playing my keyboard. i am also looking forward to having a blanket, because i am tired of being woken up at 4am by my own shivering.

speaking of beddings, i may also have my new mattress in 14 hours or so. i am excited for that. i am less excited by the logistics; if the movers aren't done by the time the mattress delivery people arrive, i don't know what's going to happen. if they double park, the electric buses (lka trackless trolleys) may have difficulty getting around, and the last thing i need is for more bus drivers to be mad at me.

when i got to work this morning, there was a paycheck sitting on my desk. it has been a long and expensive month.

tomorrow is back to the track for a brutal 6 x mile workout. after that, there's only one more track workout before the race. i should go to bed.

today was my first day at the track since i moved into my new place, and i did enjoy the fact that it's 10-15 minutes closer, not to mention being all downhill (on the way there). today's workout was great (10x800m):

1. 3:04
2. 3:03
3. 2:58*
4. 3:00
5. 3:03
6. 3:07
7. 3:09
8. 3:09
9. 3:10
10. 2:58*

(* = new PR)

(previously)

unrelatedly:

Freeing up space on our highways increases the flow of traffic and saves commuters' time, money, and gasoline.

-- Wendell Cox

ok, i call bs on that. freeing up space on highways increases the number of other people who can now drive on highways, slowing down traffic, and wasting commuters' time, money, and gasoline.

despite these lies, it is difficult for me not to be excited about being able to vote on some sort of train initiative this fall. i mean, what's the worst that could happen?

i recommend checking out the rendered stills. but the ones of anaheim suffer my typical complaint about these things: why are those people in the park? i mean, i could understand it if there were awesome shipping containers there, but if it's like a certain other park i never saw any people in there's just some smelly river and nobody lives nearby.

also i wish we could build things out of something other than glass.

9:00 I am sitting in my apartment, ready for the cable guy!
10:55 My phone rings. Cable guy is heading over; it'll take about 30-45 minutes.
11:05 I do some research on the internet, and discover comcast's on-time guarantee.*
11:10 Comcast phone rep says i should see a credit for the $100 install (!!!) on my next statement.
11:15 Cable guy shows up.
11:25 After doing who-knows-what in the basement, he comes up and says it should be working. It isn't working right away. He goes downstairs to get a modem out of his truck, and by the time he gets back the modem has a link.
11:30 After closing my firefox and system preferences, he can't figure out how to open safari (i wish i could have seen phil's face had it been his firefox that was closed). The guy is trying to type one-handed, and is fat-fingering everything. he stumbles through the wizard and i create my comcast.net account.
11:35 The wizard thing starts spewing javascript errors. We close it, and cannot get past that point in subsequent attempts. All urls are redirected to some comcast setup page (he determines this by randomly selecting a link from my history), which tries to download a .dmg image. that returns a 403 error, but of course he didn't figure that out.
11:40 I figure up Vista in a VM (should have done that in the first place), but we basically get the same results. Tech support is having problems, i don't know.
11:55 He gives up and says I should call later today or something. I sign a piece of paper that says i am satisfied with the work done, even though it is clear that i am not.
12:00 I click on the "For Technicians" link on the redirect page, and start its stupid wizard.
12:05 I am on the internet.

* GUARANTEE MAY NOT BE AN ACTUAL GUARANTEE

this was actually the first time i was home for a cable modem install, but i have learned some valuble lessons, such as do not let them touch your computer. i just rued again.

in other news, people in wheelchairs were on both of my commutes today. i guess that's what happens when you live next to a hospital.