bufferbloat

you know how stuff zooms up to incredibly high xfer rates and then just stalls.  this is an interesting journey down a series of passages all twisty, not necessarily all alike.  this stuff is not for the faint of heart, but there’s definitely some interesting work going on there.  jim gettys’ has been doing some interesting work here and while the knobs available for turning are a little arcane, following these lists will get you to poke at your OS’ networking stack in new and novel way to see what happens.

of perhaps even more interest here is the BISmark project over at GA tech where there’s some interesting instrumentation work afoot. i’ll handily step aside from the various political and service provider motivations around measurement, deriding measurement, etc..  (these machinations hit acutely close to home for me.) but i will point out that measuring the performance and the characteristics associated with a beast as unwieldy as the internets is incredibly fascinating … as an engineering problem.

LISP

i’ve been poking at LISP in some form or another for the past couple of years.  as it’s matured and is starting to see more enterprise applications and deployment i’ve been finding all sorts of novel uses for this.  if my october weren’t so packed, already i would have my ass glued to my seat for the upcoming NANOG session discussing the various map-n-encap flava’s that are floating about.  i can’t quite put my finger on the appeal here.  still, it’s fun.

fwiw – the lispmob folks just shipped a reasonably useable implementation of the mobility functions that LISP provides.  worth reading the docs and reviewing the use cases if nothing else.

DIY book scanner

when we were at the maker faire in NYC i checked out this guys stand.  he had a remarkably slick book scanner assembled and was demonstrating it.  the design is slick and the speed with which you can do the scanning is pretty impressive.  they’re talking about being able to scan 1000 pages / hour.  every library should have one of these.  there doesn’t appear to be one of these at the twin cities hacker space either.

do-it-yourself ikea

more stuff from  the maker faire.  these folks have a few designs that you can download and tweak.  subsequently, you get someone to laser cut it and you assemble it into your custom, one of a kind furniture.  bespoke household items and one-off manufacturing was a persistent theme at the maker faire, and these folks seemed to have some of the most polished stuff in the mix.

of course he nearly immediately fixed his gaze longingly upon the nearly discarded iphone 1st generations that were sitting on the shelf and he somehow persuaded me to see about jailbreaking the phone so he could use it for an upcoming trip. given my need to just have my phone work and not having the time to hack away on my phone, i have largely ignored the jailbreaking phenomenon. i simply pay for the phone that works on the network where i get support and i move along.  it’s not interesting, but the focus is the workflow.

for folks who hack on iphones as a matter of hobby, this is the mythic maze of twisty passages all alike and there’s no shortage of lore, clueful and clueless folks and tons of poseurs in this particular subculture.  however, a few googles and i’m knee deep in the world of colored sn0w jokes and DFU mode on the phone.  now here’s the rub, when you’re hacking devices which are nearly 4 years old and have subsequently gone through numerous iterations, this culture seems to leave a lot of poorly organized detritus in its wake.  folks have moved their attentions to hacking the 3G versions, the iPhone 4 and dealing with the various baseband hacks that apple has rendered useless. this involves no shortage of following scores of spotty web sites to dead ends and working one’s way through oodles of lame message boards.  in the end, the cable and a little bit of redsn0w got the job done.  followed up by a jailbreakme.com fix to correct a horked cydia installation.

some interesting tangential observations:

• there are some teenagers with some impressive technical savvy that pull together some reasonably decent screencast and tutorial videos.  i lacked the patience to wade through them, but i was surprised how google is willing to prioritize video search results and how these things seemed to have pretty high hit and watch #s.  good job guys.
• oh. my. god. is the 1st gen iphone slow, or what? we thought this shit was amazing at one point. how quickly we get spoiled.
• t-mobile coverage in our house is for shit.  if little dude wants to use this as a phone he’s going to have to go out on the deck to make his calls.  i might be forced to reconsider how crappy i consider AT&T’s coverage to be.

the RFC while largely tongue-in-cheek in intent as an april fools RFC has repeatedly proven itself to be rock solid philosophically.   who knew that a mere …

zendoggy[sulrich]% wc Sites/rfc/rfc1925.txt
     171     576    4294 Sites/rfc/rfc1925.txt

ahem, 576 word RFC would be chock full of so many koans.  re-read this one often.

then i tried mint, it was good, but i couldn’t enter transactions manually, the projection and cash flow tools were poor and the budgeting tools were horrid.  but, it was free and if you could come to terms with the faustian bargain that was having a 3rd party collect financial information on you, it was reasonably workable.  then they got big, their feeds and their back end started to suffer, the customer service interaction was pretty horrid.  WTF – forums?  i know that’s very web 2.0 but i need to get resolution to these issues, not fuck around in a forum commiserating with a bunch if like-irritated customers.  if i give you money to provide me with a reliable, secure and supported service will you hook me up?  there’s a lot to love about what you’ve got going there.

then quicken/intuit acquired them and the community puckered up their bungholes.  quicken for the mac had been a train wreck, quickbooks is an application that pretty much everybody loathes but there really don’t appear to be better options so everyone just sucks it up and deals.  but, there was a contrite blog posting about how mint would be the same old mint it ever was (talk about a mixed message) and this would provide the infusion of cash to make the enhancements, etc. that we’d all been clamoring for, etc. then they added goals,  nice.  they improved a number of things but you still had the anemic customer care / support and it seemed to require a lot more hand holding than you would expect something along these lines to require.

we plugged along then back in feb / mar time-frame, they made some sort of unholy change to the back end and all hell broke loose.  i had dead accounts which couldn’t log in, i was asked to duplicate things, merges were horribly horked up, etc. i gritted my teeth and worked through these issues.  this necessitated periodically dealing with erratic duplicate transaction explosions which threw off budget tracking and a host of other things.

then about a month ago i did a little cleanup, following the directions that mint helpfully provided. which i found irritating given that apparently there were enough people experiencing these problems that they had to develop documentation to facilitate manual cleanup on the part of their users.  how messed up is that?  seriously, you make back end changes to your system and you provide your customers with instructions to address the scorched earth situation that  you’ve created?  WTF?

following this “clean up”, i was left with a collection of accounts for which i could apparently get budgeting and expense information but i could not see the actual transactions within the accounts.  double WTF?  i did what you told me to do and you don’t even give me the courtesy of a new opportunity to clean up some freshly created mess?  you just hide a bunch of stuff i need, from me, your apparent bitch.  well, it’s a free service so it’s not like i’m out any money, just a lot of irritation and time.

so here’s the recurring punchline – i’m quite willing to spend a reasonable amount of money for a solution that works.  mint had a good chunk of the stuff that was useful to my household and they were pretty bright about sending me reasonable information on optimizing my finances and usage.  i would happily pay a monthly fee for online tools that work and straddle the various accounts that i have spread hither and yon.

now let’s talk about desktop software packages.

after trolling through various software reviews and looking at inscrutable screenshots, perusing feature lists and getting my wife’s input we settled on kicking the tires on a few packages …

MoneyDance 2011 for the mac – sweet they have a demo i can try.  crap,  it’s limited to 100 transactions. oh, and they’ve disabled the online banking integration elements.  shit. next.

iBank 4 – great a 30 day non-transaction limited demo.  seems intuitive enough, handy tool to check out the supported financial institutions for direct access. looks like the interesting banks, brokerages, etc, are supported, sweet.  start adding them. login failure, are you sure?  what?  wait a second here, seriously?  try again, INBOX blinks. too many login failures? account locked? oh hell.

the balance of the options out there look like crap. i’m loathe to wade through that morass. best make one of these work.

update – 16-july, 2011

it’s worth pointing out that we did in fact settle upon iBank.  i worked through a handful of the foibles that i had.  its ability to interact directly with the bank in terms of getting transactions, etc.  the trial leaves some of the annoying warning dialog boxes in play despite your request to not keep showing them in the future.  this is addressed in the production paid-for version.

the process of getting all of your financial information from these systems and understanding your cost-basis, etc. is largely subject to the whims of your financial institution and their willingness to give you access to all of that information.  jp morgan chase, you can bite me. 90 days of downloadable history is shamefully little history.

we’ve actually been able to get access to information via iBank that we couldn’t get via mint and this has proven to be a welcome addition.  the iphone app is pretty cheesy.  but it’s useful for periodically checking what you’re up to and manual transaction entry is handy.  things are looking up on this front, not great, but not bad either.

oddly enough this just seems to be  a regurgitating of the same old links on different sites and repetitive pontificating on italian designers and drooling over look books.   the silver lining in all of this is at least a marked reduction in douche driven fashion and at least a revived interest in quality over quantity.

i’m kind of curious to see how long this particular fashion trend will last.  for those of you concerned that this will mean you won’t be able to distinguish yourself through the exorbitant expenditure of cash, worry not, these guys still seem to fetishize all sorts of unreasonably expensive stuff.

ld: huh? what’s a chalkboard?

me: seriously? what’s a chalkboard? oh well – we can file that under sayings which are increasingly irrelevant.

sorry that should probably be “flirty-girl style”. (hyphenated)

wtf?

nice.

setting aside that bit of silliness on my part, i noticed something. while i was grinding out real work and answering emails between the discovery of the breakage and repair, i discovered that my work flow had developed a dependency on having an automated “checker” come back and clean up after me.

now, i’ve had this realization in the world of MSFT Word and other packages with built-in spell checkers, but clean-up of typos was one of those things that i did in a batch manner, my fingers hadn’t actually adapted to the use of an integrated spell checking function like they had in my normal work tools. (read, emacs)

i’ve always had some level of pride in having relatively decent spelling skills. but the power of ctr-; and knowing what your tool would do, has had an interesting impact on my productivity. one that i’m not entirely convinced is for the best in the long term.

“i made that pirate ship.” “oh? what did you make it out of?” “playmobile.”

apparently playmobile is a raw material these days.

$Id: osx-nits.txt,v 1.4 2001/05/03 11:50:52 sulrich Exp sulrich $

# os x nits
## general os and interface nits

 * when i hit the cmd key in a modal dialog box - i'm not presented with a
   list of keyboard short-cuts for the dialog box. it's a capital letter
   crap-shoot!  it would be nice to be able to hit the cmd key and have the
   corresponding keyboard shortcut displayed.

   >> it bears noting that this interface nit still stands.  however,
   >> i've come to terms with this to some extent.  changing the
   >> extension on a file from the finder is no where near as annoying
   >> as it used to be.``

 * it would be nice to have a real print manager that worked. the current
   printer controller is really lame and doesn't allow you to view the
   configuration of current devices.  it looks like the menu item is there
   but it's grayed out.

   * further to gripe about the printer support there appears to be a lack
     of real definitions for printer description files that you install.
     meaningful names for the printer definitions that are there would be
     really nice.

  >> this is an area where i have to say apple really stood up and did a
  >> good job. i admittedly have rather modest printing requirements,
  >> however, for a guy that has to add printers in foreign offices at
  >> work all the time, the support is wide and the hassle is minimal.

 * what i wouldn't give to be able to remap the apple key in terminal to
   behave like a meta key.

   >> this gripe stands. i've ameliorated this to a large extent through
   >> the use of the shipping X11 implementation and the use of
   >> xterm/rxvt. still, this leaves me with a nagging issue wrt cut and
   >> paste.

   just a follow up to this - you can kind of do this in emacs for mac os x
   with the option key but that is just a little too awkwardly placed for
   my tastes.  looks like i'll be an escape-key emacs user for a while

 * why must changing the desktop background color be such a chore? if i want a
   solid color why can't i just select one w/the color tool? i surmise that i'm
   missing out on something fundamental here but it's really annoying

   >> this gripe still stands. sure, there's a solid color option there,
   >> but you still don't get the color tool.

 * teeny hardware nit here.  why is the escape key so small?  the ~ is
   in the right place though and this is nice.

 * virtual desktops! - i'm running out of screen space and there is no
   virtual desktop functionality that i can find anywhere.

   >> spaces. need i say more? sure spaces isn't perfect, but it's
   >> pretty damn close and it's built-in with reasonable defaults.
   >> color me tickled.

 * make the PDF generation access a little more prominent on the print
   dialog.  while i love the ability to have PDF output i hate having to
   select that option repeatedly.  i'd like to make pdf output my default.

   >> well, you still can't make PDF generation the default, (to my
   >> knowledge.)  but it's featured much more prominently and there's
   >> slick integration that's taken place with the ability to direct
   >> output to applications.  which has my Yep! setup quite happy.

## terminal
 * terminal performance over the long haul is horrible.  this needs to be
   sped up considerably.

   >> fixed this, but i still don't use terminal - ref gripe re: cmd key
   >> utilization and emacs keybindings

 * more on terminal.app - this needs some real work on the termcap side of
   things.  there doesn't' seem to be a really consistent mechanism for
   making sure that you're dealing with the terminal on this.  what i
   wouldn't give for just really decent xterm support.

## airport / wireless
 * how about some meaningful stats regarding signal strength?

   >> still sucks if you ask me.

 * opening up the driver interface on the airport cards to support things
   like airsnort.

   >> improved considerably. there's a pretty wide range of support here
   >> and folks have filled in a lot of the gaps relative to tools for
   >> sniffing out additional wireless APs, etc.

## general networking
 * IPv6 support (addendum - looks like this will be in jaguar)

   >> pretty damn good support if you ask me. 

 * multicast support (w/support for IGMPv3)

   >> grrr. still outstanding. 

 * i would really like dummynet or some form of queueing for network
   traffic.  this would let me use my mac for testing some network
   configurations. altq would be heaven.

   >> no progress here.  understandably, this is a consumer OS after
   >> all.

 * pf would be preferred vs. ipfw (in all honesty i haven't had a chance
   to see if the necessary kernel hooks are there for pf)

   >> there's been a lot of forward progress on the firewall front. i've
   >> added to this with the use of little snitch which provides a lot
   >> more granularity in terms of the per-application visibility that
   >> you'd be interested in. still. pf is pretty much the bomb for this
   >> stuff.

overall, i really have to say a lot of my longstanding nits have been admirably addressed. if you look at this from the perspective that there are a fair number of nits that are purely networking dweeb oriented in nature, and you keep in mind that this really is a consumer oriented OS, you have to be impressed at how well they’ve addressed the spectrum of user requirements.