<wolfspraul>
wpwrak: Adam is about to order a few more pieces of this and that chip. Do you have some final idea about the best reset ic already?
<aw_>
how about the new reset ic?
<wolfspraul>
it turns out we already sourced 100 of the 4.4v reset ic under consideration now :-)
<wolfspraul>
but no hesitation, if we think another one is better, then those 100 will proudly go to the pile of stuff we left behind in our search of perfection...
<wolfspraul>
but if Adam would know which one it is, he could combine the purchase with other stuff and we don't have to rush buying chips at last minute
<wpwrak>
hmm no, no final recommendation yet
<wpwrak>
the 4.4 V chip seems to behave well at least when the system is up. need to check how it performs with a "cold" system and then plugging/unplugging things.
<roh>
freezer tests again?
<wpwrak>
naw, i think the freezer theory needs filing next to the aether theory :)
<wpwrak>
the bizarre change in corruption rate was probably just the uncommanded locking i discovered later
<aw_>
wpwrak, will you rise environmental temperature and cool down to test 4.4 V chip repeatedly?
<wpwrak>
aw_: no, i don't plan to include temperature changes in my measurements. too complex for my infrastructure.
<wpwrak>
but i'm thinking of varying the supply voltage, e.g., to 5 V +/- 0.3 V
<wpwrak>
but fist i want to kill that midi hang. alas, it's MIDI xor NOR bugs :)
<wpwrak>
when to you expect to send M1rc4 to the layout folks ?
<wolfspraul>
sure no rush
<wolfspraul>
we are loaded with good productive work everywhere
<wolfspraul>
only unpaid unfortunatelly :-)
<wolfspraul>
no no
<wolfspraul>
no expect
<wolfspraul>
tons of work
<wpwrak>
;-)
<wolfspraul>
just sold another 2 m1 last night :-)
<wolfspraul>
adam got the yield to 76 now, super great I think
<wolfspraul>
considering how it started
<wpwrak>
so when will you run out of M1rc3 ? :)
<wolfspraul>
hmm
<wolfspraul>
individual sales are slow
<wolfspraul>
basically comes down to finding friends
<wolfspraul>
then there is an occasional 5-pack or 10-pack, those are unpredictable
<wolfspraul>
I don't know
<wolfspraul>
roh: do you have some friends you can talk into buying an m1? :-)
<wolfspraul>
I haven't looked at numbers recently, but sales are probably a little less than 40 now, and about 35 in stock
<wpwrak>
so you may still run out this year
<wpwrak>
maybe we need a christmas-themed patch, to boost seasonal sales :)
<wolfspraul>
he
<wolfspraul>
of course I may run out, I hope so
<wolfspraul>
the key is telling the story right
<wpwrak>
and tell it to the right people :)
<wolfspraul>
definitely
<roh>
heh.. currently none with money who are in the vjin business
<roh>
+g
<wpwrak>
lekernel: btw, have you tried to reproduce the hang ? the following should work: connect a MIDI device with slider. bring up MIDI settings. jerk the slider back and forth as fast as you can. after a few seconds (sometimes just 1-2), the M1 will stall
<wpwrak>
wolfspraul: btw, are you following up with M1 customers to see what they're doing with it ?
<wpwrak>
and ... do we know the evolution over time of the number of subscribers on milkymist-devel ?
<wolfspraul>
he
<lekernel>
wpwrak, I have used MIDI sliders a lot of times without problems. that was before your patch though.
<lekernel>
the subscriber count is stagnating, there's perhaps a new one every 6 weeks
<wpwrak>
yes, the patch makes the hang happen nearly instantly, at least in the MIDI settings dialog
<lekernel>
right now it's 109
<wpwrak>
hmm, so customers don't turn into subscribers. perhaps the "devel" scares end users (if we already have touched that segment)
<wpwrak>
wolfspraul: when you ship an M1, do you also tell people where to go for more information ? or are you sure they generally know already ?
<lekernel>
yeah, I'm not complaining about such issues for no reason :(
<wpwrak>
i think we should consider the M1 project in its present state as still fairly immature. what has basically happened so far is the trailblazing necessary for a proof of concept. basically implementing one use case. that alone is a wonderful achievement, but of little practical value to most people.
<wpwrak>
i.e., only those who can stay very close to that use case would actually have a "rounded" product. while the others will find a lot of things missing, be it features, stability, or even community structure
<wpwrak>
but i think the slow growth we're seeing isn't necessarily a bad thing. it would be nice if it was a bit faster, but there is some buildup that still needs to happen before you can successfully approach regular customers
<wpwrak>
where i'd define "successful" not only as "make them buy" but also "keep them happy after they've bought it"
<wolfspraul>
where to start
<wolfspraul>
no I don't follow up
<wolfspraul>
we are lacking efficient means to do so
<wolfspraul>
mailing lists themselves are stagnant, that's not a way to reach any non-tech audience
<wpwrak>
following up would be a good start :) you could just send them a mail a few weeks after the purchase. asking if everything is well, if they've run into problems, etc.
<wolfspraul>
and yes, the product is far from perfect, but the key is to keep fixing and keep being nice to people that respond anything back
<wolfspraul>
sure I could
<wolfspraul>
but like I said we are lacking efficient means to do so
<wpwrak>
most people should like such personal post-sale attention, and not see it as spam/annoyance
<wolfspraul>
nah
<lekernel>
I tried to open a forum once, it wasn't quite successful
<wolfspraul>
29 units went to people who will resell
<wpwrak>
(efficient means) why ? write a standard letter and send it to those ~50 M1 owners. is that too hard ? :)
<wolfspraul>
communication through them is a pain
<lekernel>
I think there was exactly one person who posted there in 2 weeks, and a hundred spam robots
<lekernel>
at the end PhpBB, this piece of shit, got hacked (even though I had the latest version and security patches) and I closed the forum
<wpwrak>
yeah, a forum currently has the problem of small numbers
<wolfspraul>
the product can and will get much better, and that will attract more users
<wolfspraul>
that's the only way I know and motivated to work on
<wolfspraul>
actually
<wolfspraul>
lemme send out the faderfox reply :) (cc werner, all in German though so I cannot really cc Sebastien)
<wpwrak>
sure. but we should improve the communication beyond the "inner circle"
<lekernel>
cc me anyway, I can manage to translate
<lekernel>
and I need to learn German at some point
<wpwrak>
wolfspraul: how many resellers in total ? and maybe you can ask them to either pass your the e-mail of M1 buyers, or if they don't want to do this, forward your mail ?
<wolfspraul>
he ;-)
<wolfspraul>
sometimes you seem to live in some far-away paradise world
<wolfspraul>
maybe that's the Buenos Aires effect
<wolfspraul>
I have cc'ed Sebastien a few times on mails here and there, so I save myself from him asking the same questions :-)
<wolfspraul>
not everybody is as excited and focused on Milkymist as we are
<wolfspraul>
they just don't reply
<wolfspraul>
they don't care
<wolfspraul>
they reply 2 weeks later
<wolfspraul>
or 4 weeks later
<wolfspraul>
they reply to 1/5th of what I asked
<wolfspraul>
and so on
<wolfspraul>
I can make this my full-time job to force answers out of people who don't care, but that's a real waste of time and I won't do it
<wolfspraul>
the key is to find more people who care, and become proactive *themselves*
<wolfspraul>
new people
<wolfspraul>
the old one already know us, and do what they do
<wpwrak>
the objective isn't to force answers, but to open a channel
<wolfspraul>
the pull has to be on the other side
<wpwrak>
and maybe find out a bit more about what people do with it
<wolfspraul>
:-)
<wolfspraul>
you don't give up easily
<wolfspraul>
we know that :-)
<wolfspraul>
how about: nothing
<wolfspraul>
:-)
<wolfspraul>
(it is slowly improving, but still true to a meaningful extent)
<wpwrak>
where should the pull go ? to the reseller ? to the sharism online shop ?
<wolfspraul>
it's easy to find the milkymist twitter feed
<wolfspraul>
I think that comes natural for chatting nowadays
<wpwrak>
if you tell people about the irc channel and the devel list, many may be a bit lost. the devel list may be too technical for them. the irc channel, too. plus, hanging out on irc 24/7 isn't for everyone
<wpwrak>
so they don't communicate and may even think it's their own fault -> nothing happens
<wpwrak>
(twitter) hmm, maybe. that's a world i know nothing about :)
<wpwrak>
if you approach the customers and make them know that you're willing to listen to their problems and questions, then they know already one point where they're welcome and that isn't too hard to reach
<wolfspraul>
I have seen some positive reactions lately
<wpwrak>
if you don't have the time, that's harder, of course
<wolfspraul>
let's hope they don't fizzle out
<wolfspraul>
for example n0carrier who showed up here :-)
<wolfspraul>
was that just the excitemend of the first day?
<wolfspraul>
and his m1 is proudly retired to a drawer now?
<wolfspraul>
or will he continue?
<wolfspraul>
those things are invaluable, and cannot be enforced
<wolfspraul>
I hope he continues to use it
<wpwrak>
let's see. he said he had lots of trips lines up
<wolfspraul>
I hope he shares awesome patch improvements back
<wolfspraul>
and his experience with controllers
<wpwrak>
and yes, i'd love to have more nocarriers around :)
<wolfspraul>
and videos
<wolfspraul>
and and and
<wolfspraul>
but... maybe nothing?
<wpwrak>
exactly
<wolfspraul>
I don't know
<wolfspraul>
he is free
<wolfspraul>
we sold one to the lzxindustries.net guys
<wolfspraul>
interesting stuff they do
<wolfspraul>
why do they buy m1?
<wolfspraul>
just to check it out, then drawer?
<wolfspraul>
or more?
<wolfspraul>
don't know
<wpwrak>
there are two things that need to happen in communication with customers/users: 1) find out where they want the product to go 2) inform them better what it can do
<wolfspraul>
I don't disagree with you
<wolfspraul>
I plan to spend time talking to people one by one
<wolfspraul>
like the faderfox guy you found
<wolfspraul>
just sent the reply btw
<wpwrak>
e.g., none of the existing patches use MIDI controllers. a few use DMX in, at least one DMX out. so mosr customers may not realize that they could greatly improve their M1 performance by adding a controller
<wolfspraul>
agree
<wolfspraul>
better documentation, better handbook, better patches, better update mechanism (better communicated and regular updates and news about updates), etc. etc.
<wolfspraul>
we know it all, just getting to do it step by step
<wolfspraul>
we sent one unit for review to a UK distributor, let's see what happens there
<wpwrak>
lzx .. ah, see what the competition is doing :)
<wolfspraul>
no competition - just friends
<wolfspraul>
I think M1 is extremely unique
<wolfspraul>
something like this just will not be developed or marketed by anyone
<wolfspraul>
it's too risky, too expensive, too exotic
<wolfspraul>
(too expensive not in the sale price, but in development cost for a regular company)
<wolfspraul>
so we need to see what kinds of market positions we can get into that simply nobody else will be willing to match
<wolfspraul>
some unique features, unique ways of combining things. I think we are close...
<lekernel>
Alles offen aber alles buggy?
<wolfspraul>
sure
<wolfspraul>
you will get used to my humor eventually :-)
<wolfspraul>
that's what he said, so I just share his view. let me translate his version, it's more diplomatic
<wolfspraul>
"the way I see it your project is of course wide open with almost infinite possibilities, but not really very user friendly yet in terms of customers who do not necessarily want to dive deep into programming"
<wolfspraul>
ha ha
<wolfspraul>
isn't that great!
<wolfspraul>
I love it
<wolfspraul>
so I shortened that to "everything open but everything buggy"
<lekernel>
we really need to get out of this perception...
<lekernel>
developers won't buy them because they are scared of the complexity
<lekernel>
and users won't buy them because they think it's a programmer thing :-)
<wolfspraul>
yes
<wolfspraul>
and when he (or me in that case) say "buggy" that's not necessarily a technical bug only
<wolfspraul>
it can also be lack of documentation
<wolfspraul>
like "what controllers work with m1?" (werner above)
<wolfspraul>
but I don't want to point fingers, I feel the todo list on my side very well and glad to work on it step by step
<wpwrak>
that pretty much corresponds to my experience as well :) plug in audio for the first time -> boom. connect my RF keyboard -> mouse pad doesn't work. get serious with MIDI -> message loss (and hang). of course, not EVERYTHING is buggy. but a lot (or perceived as such). as is to be expected for a complex product that has gone through very few hands so far.
<wolfspraul>
I think most tech products are loaded with bugs. if we can really make a product one day that actually works in most user environments - WOW!
<wolfspraul>
I guarantee you it will sell like hot cakes.
<wpwrak>
but of course, the more end-user your customers are, the less they will understand such shortcomings
<wolfspraul>
yes agree [werner]
<wolfspraul>
but we are geting better
<wolfspraul>
getting
<wpwrak>
also the milkdrop language is mildly horrifying. i perceive it as a bit of an insult to the IT profession :)
<wolfspraul>
I looked at amazon comments for korg kaossilator pro
<wolfspraul>
and what's the first thing I see "serious usb-midi sync bugs"
<wpwrak>
yes, M1 needs to get a bit more rounded. meet "normal" expectations of customers. many things are at a "proof of concept" stage at the moment
<wpwrak>
heh ;-)
<wolfspraul>
let's assume this is a genuine user, am I surprised? no way
<wolfspraul>
these types of products are crazy hard to "make work well"
<wolfspraul>
and Korg is a very respected brand doing this for 30 years
<wpwrak>
lemme check this out :) that's may be something i found, too
<lekernel>
there are tens of thousands presets which have been contributed to MD, no matter the "horrifying" language
<wolfspraul>
so one by one. We cannot expect perfection today, and I will as much openly talk about the shortcomings as I praise the things that work.
<wpwrak>
yes, yes, given enough thrust, pigs fly rather nicely :-)
<wolfspraul>
I bought this Optoma projector
<wolfspraul>
another big brand
<wolfspraul>
and... believe it or not. they didn't manage to auto-detect the presence of a VGA signal.
<wpwrak>
(midi sync on the kaossilator) ah no, that seems to be something else. something i wouldn't have run into
<wolfspraul>
so when I plug in VGA, or just reset the M1, I have to fiddle through a bunch of menus to make the VGA show up again.
<wolfspraul>
and that's for a projector!!!
<wpwrak>
(crazy hard) yes, MIDI is a communication protocol. there are always two sides to it :)
<wolfspraul>
whose only purpose is to *project*
<wolfspraul>
and a company that is selling this projector at least in tens of thousands of units is not able to get the vga signal detection to work?
<wpwrak>
;-))
<wolfspraul>
that's the reality in the tech industry
<wolfspraul>
everybody knows
<wolfspraul>
and that's why the lines on an Apple launch date are so long
<wolfspraul>
because people are fed up with it
<wolfspraul>
every time I reset my m1 ... click click click, etc.
<wolfspraul>
I hate Optoma!!!
<wolfspraul>
and those buttons don't even have good tactile feedback, but whatever...
<wolfspraul>
so I am calm on M1
<wolfspraul>
we need honest feedback, then fix bugs, update, go at it again
<wpwrak>
yup. that's why i'm suggesting to bring up that feedback channel.
<wpwrak>
qi-hw is already operating a forum. has that one been a major maintenance burden ? if now, would it be possible to reuse the existing infrastructure for milkymist ?
<wolfspraul>
that forum is not successful, using mediawiki liquidthreads
<wolfspraul>
maybe I try phpbb one day, the most commonly used forum software
<wpwrak>
you mean it's too hard to use ?
<lekernel>
we also need to improve big time on communication/stories/PR, as I said many times this Broadcom project has incredibly more popularity than MM without shipping a single device
<wolfspraul>
liquidthreads has not been a maintenance burden, no. but also people are too unfamiliar with it I think.
<wpwrak>
hmm yes, looks a bit weird
<wpwrak>
everything on one page ? who need scalability ? :)
<wpwrak>
also, apparently no linear view
<wpwrak>
okay. not the best way to welcome newcomers
<lekernel>
when this broadcom project started, their website was merely a text file and a few pictures. yet, it gathered in 2 days way more media attention than MM in 4 years
<wpwrak>
what is "this broadcom project" ?
<lekernel>
raspberry pi
<wpwrak>
aah. well, it's cute, cheap, and broadcom is a big company.
<wpwrak>
(though i didn't even know they were behind it)
<lekernel>
wolfspraul, I can recommend you against phpbb. it's a massive security problem; the one I installed got totally hacked in 4 weeks, and even in the meantime spam bots were finding ways around the captcha
<roh>
phpbb? bwahahahah
<wpwrak>
so what does the CCC's forum run on ? ;-)
<roh>
we dont run any. simple.
<wpwrak>
so what would the CCC's forum run on ?
<roh>
there wouldnt be one.
<wpwrak>
thinking of it, i should still have some telnet-based forum software sitting around somewhere. that one never got hacked :)
<lekernel>
punbb looks a bit saner
<roh>
forums only work if 'knowning' people read them and answer questions. they are bad for storing 'knowledge' and force inferior userinterfaces onto people. see ubuntu. you either get proper information in wikis/dokumentation pages or you read 'user guesses' in formus
<roh>
eh forums.. the latter always being the informations of the lowest quality
<wpwrak>
forums work quite well as an information source if google indexes them
<roh>
wpwrak: only if they get filled by people who KNOW answers.
<wpwrak>
i find most of the answers to technical issues in fora
<wpwrak>
and of course, someone has to provide answers ;-)
<wpwrak>
but dropping off an answer is easier than maintaining a wiki page
<wpwrak>
most wiki pages are "write and forget" anyway
<wpwrak>
http://www.go-mono.com/forums/ has the same. to use the "back" button, you have to click it twice in rapid succession
<wpwrak>
ads inside threads (just one per thread on netty, it seems. several on go-mono)
<wpwrak>
so .. mixed impression.
<Fallenou>
just to let you know I presented shortly Milkymist at the OSHCamp in London on thursday the 27 of october, a few interested guys came to me afterward to get some more intel, I did my best to describe the project to them, they seemed interested
<wpwrak>
kewl ! we need more developers ! developers ! developers ! ;-)
<Fallenou>
( developers ! developers ! developers ! ) < isn't this a quote from steve ballmer ? :p
<wpwrak>
of course it is ;)
<kristianpaul>
(milkdrop language is mildly horrifying) YES YES :-)
<Fallenou>
it's a first draft of what I think could be a pipeline example
<kristianpaul>
is okay, really i mean is documented, you wonder how many people got what is a pipeline in hw thanks tou you :)
<Fallenou>
I am trying to get together a few elements in order to build a small 8-bit cpu, it's only purpose will be for me (and anyone reading it maybe) to learn about how it works
<Fallenou>
thanks !
<Fallenou>
will have a hard time adding support for a new arch to llvm,
<Fallenou>
is reading llvm-lm32
<kristianpaul>
if works for you and is well documented (both hdl code and datasheets) worth the effort i think
<kristianpaul>
Anyway have fun !
<Fallenou>
yes I hope :)
<Fallenou>
thanks !
<Fallenou>
will keep posting about it anyway
<Fallenou>
Instruction Pipeline is under construction
<Fallenou>
anyway, for those wondering, it's not because I speak less here that I don't read everything which is said down here.
<Fallenou>
I keep reading ML as well
<lekernel>
yeah LLVM backends are a bit messy too
<lekernel>
nowhere near as bad as GCC, but still messy
<Fallenou>
but I guess it's better than gcc
<Fallenou>
yep
<Fallenou>
I havn't answered yet on the ML but I am very interested in a "Minimist"
<kristianpaul>
put sdcc over the table
<Fallenou>
a $99 product would be very cool :)
<kristianpaul>
wich had improved a bit in last versions
<Fallenou>
the only trouble I could see is dividing effort
<Fallenou>
two projects could be hard to develop in parallel and to maintain
<Fallenou>
different bitstreams/bios/code etc etc
<Fallenou>
but if it can increase the "early adopters" number, and therefore the number of hacker/contributors
<Fallenou>
it would be really good :)
<kristianpaul>
bios may be not, actually i remeber M1 bios have support for to ac97 chips already isnt?
<lekernel>
isnt
<kristianpaul>
ah
<kristianpaul>
s/to/tow
<Fallenou>
but well, if you do one minimist for $99, I will definitely buy one
<Fallenou>
count me in
<kristianpaul>
was considering buya LX9 microboard ;)
<lekernel>
well, first, from now on, I will not do anything that needs to be sold before a successful vaporware media campaign
<kristianpaul>
he, so the 99$ board is a good start campaign :-)
<lekernel>
as long as it doesn't show in google trends, I will not do it.
<Fallenou>
lekernel: a campaign for what already exist ? the M1 ?
<Fallenou>
or a campaign for the new product
<Fallenou>
gotta go see ya
<lekernel>
for any new product
<Fallenou>
ok
<kristianpaul>
bye
<kristianpaul>
so milkymist dont show in google trends, project is over now? :-)
<wolfspraul>
ah I keep talking to the wall
<wolfspraul>
your "successful vaporware media campaign" means that you raise expectations, that's all
<wolfspraul>
raising expectations is a pretty good way to fail, depending on what your actual goal is (maybe the intermediate publicity is all you need to land a new job, that's another case)
<wolfspraul>
I hope you still enjoy Milkymist and M1, and keep hacking on it. You definitely cannot complain about "nothing to do" :-)
<wolfspraul>
I think we are getting closer to establishing Milkymist as a serious technology and product, but 'closer' is not enough.
<wolfspraul>
what M1 needs next are serious large buyers
<wolfspraul>
someone who thinks this is actually a great product and orders dozens or hundreds
<wolfspraul>
what we ship today is not at the quality level such a buyer would need
<wolfspraul>
not yet
<kristianpaul>
likes its M1 and ignore google trends
<wolfspraul>
and... back to work :-) you keep envying Raspberry, I don't, I just hope they are successful.
<kristianpaul>
heads to pasto
<wolfspraul>
nice
<wolfspraul>
wolfgang heads to coffee
<wolfspraul>
Fallenou: why do you think a 99 USD product would be cool?
<wolfspraul>
you want to buy an m1 but you don't have enoug money?
<wolfspraul>
enough
<wpwrak>
(vaporware) yeah, IPv6 is a great example of how that backfired
<wpwrak>
better have the thing 100% ready by the time you heat up the hype
<wpwrak>
wolfspraul: M1 at USD 99 and still profitable would definitely be cool. such would be the ability to fly to distant stars in the blink of an eye, teleport down to planets, always breathe the air there, and converse with the local humanoids in fluent english
<wolfspraul>
:-)
<wolfspraul>
let's see whether we can help Fallenou first