<wpwrak>
wolfspraul: the directory thing is a little weird. i thought i had once submitted a patch that fixed it but i'm not sure anymore. maybe there was also a regression.
<wolfspraul>
yes. :-) next time you choose a higher version number I should be able to rebuild the package in a day...
<wolfspraul>
wpwrak: for boom, we just had a case in Milkymist One where the vendor increased the acceptable voltage range of a part without changing the part number
<wpwrak>
maybe the old ones were good for higher voltage as well ?
<wolfspraul>
in general it's probably imaginable that not all criteria that are required for a project will trigger a part number change on the other side (manufacturer) when changed
<wolfspraul>
say for example some mechanical property is actually required, but not expressed in the part number
<wolfspraul>
6 months later the mechanical property changes, without part number change
<wolfspraul>
this is a hypothetical example but I think it's a case to keep in mind
<wpwrak>
i still marvel at the stunt samsung pulled off with the s3c2442 A-to-B transition, where all the pull-ups became pull-downs.
<wolfspraul>
well what I am talking about now is that the part number may not give 100% accuracy
<wolfspraul>
because we may depend on some parameter that will not trigger a p/n change even when this parameter is changed on the other side
<wpwrak>
yes. sometimes you also have revisions. a0 has this errata, b0 not.
<wpwrak>
or not revision but date code. there's usually *something* that tells you want you have.
<wolfspraul>
this case here, the voltage range was extended to lower
<wolfspraul>
it used to be 4.5v to 5.5v, and then it was increased to 2.7v to 5.5v without p/n change
<wpwrak>
if two identically marked parts are still different, just treat them as the worst case of both :)
<wolfspraul>
if your design tries to use this component with 3.0v you better make sure to actually get the newer chips
<wpwrak>
because if you can't tell them apart, nobody else can either
<wolfspraul>
I'm not saying it's not determinable at all.
<wolfspraul>
but the p/n may not carry enough precision
<wpwrak>
(extended voltage) but are thepackage markings the same ?
<wolfspraul>
the part number did not change
<wolfspraul>
I'm sure there is some other way to tell the difference.
<wpwrak>
you could include the other stuff (revision, date code, etc.) in the part number as well.
<wpwrak>
for revision, this is not too uncommon
<wolfspraul>
who is 'you'?
<wolfspraul>
you mean in our bom?
<wolfspraul>
at some point you will always have a translation
<wpwrak>
you can be the distributor. or you the developer
<wolfspraul>
and without manufacturer part number change, the chances that the digikey/mouser/any distributor number changes is also small
<wpwrak>
translations are fine. .equ files are your friends ;-)
<wpwrak>
i think digikey sometimes have the revision in the part number
<wolfspraul>
many things are encoded in the part number, so the manufacturer will typically issue a new one for all sorts of reasons
<wolfspraul>
but maybe still not for all reasons that you would care about
<wpwrak>
significant silicon revisions are pretty common
<wolfspraul>
ok just thought I mention this case
<wpwrak>
yes, it's something to keep in mind
<wolfspraul>
maybe that's why extensive notes/comments fields are a good idea
<wolfspraul>
I remember you said the onces you saw from FIC were quite detailed
<wolfspraul>
there can always be a mismatch in criteria on what someone considers 'different'
<wpwrak>
a "generalized notes" item could be useful, yes. perhaps also for dta asheet urls and such
<wpwrak>
btw, are you trying to "boomify" a project ?
<wolfspraul>
sure I will
<wolfspraul>
mmone-jtag-serial-cable
<wpwrak>
aha ! that's why you're in a hurry :)
<mstevens>
/quit
<wpwrak>
how far along is it ?
<wolfspraul>
in which sense?
<wolfspraul>
Yanjun Luo has made 4 functioning boards
<wolfspraul>
I have sent 3 to Sebastien (not arrived yet)
<wpwrak>
so ready for MP ?
<wolfspraul>
mp
<wolfspraul>
sounds like world domination
<wpwrak>
it kinda is :)
<wolfspraul>
I need to make another 30 or 40, yes
<methril_work>
morning
<wolfspraul>
but it's also a test project
<wpwrak>
btw, who is yanjou lou ?
<wolfspraul>
so using this opportunity I do many things that will be useful later
<methril_work>
wow!! wolfspraul tell us his plans!!! ;)
<wpwrak>
yanjun
<wolfspraul>
an fpga engineering working for an American router company in Beijing
<wpwrak>
schematic are in kicad ?
<wolfspraul>
yes
<methril_work>
<wolfspraul> sounds like world domination
<wolfspraul>
layout too
<wolfspraul>
just not very clean
<wolfspraul>
I am cleaning it up now.
<wpwrak>
ah, i have it here ... having a peek
<wpwrak>
is my pcbnew crazy or did someone really copy all the component values on the side ? ;-))
<wolfspraul>
there is a lot of cleanup to do
<wolfspraul>
also some things are not committed
<wolfspraul>
wait a bit
<wolfspraul>
I was planning to ask you to have a look, but a little later
<wolfspraul>
close your eyes, reopen in a few hours
<wpwrak>
sure, no rush :)
<wpwrak>
they didn't use fped, did they ?
<wolfspraul>
no
<wolfspraul>
this project is a very nice opportunity for me to learn
<wpwrak>
pity. i'd love to see an industrially made board with fped footprints.
<wolfspraul>
I can boomify it, and fpedize as well (though later)
<wpwrak>
(as a verification)
<wolfspraul>
industrially made?
<wolfspraul>
if I do this myself, it's as much hobbyist as you can imagine
<wolfspraul>
probably if I screw up I will hand it over to Adam last minute :-)
<wpwrak>
with all the special layers. solder mark, silk screen, solder paste layer
<wpwrak>
solder maSk
<wolfspraul>
wait until we get to xue
<wpwrak>
the relatively simple fpga board would be an opportunity to validate the stuff. so far, afaik, no "full" board has been made with fped.
<wolfspraul>
which board do you mean?
<wolfspraul>
'relatively simple fpga board'?
<wpwrak>
it works perfectly for my diy stuff, but i don't know yet if i didn't screw up elsewhere
<wpwrak>
err, jtag not fpga :)
<wolfspraul>
ok
<wolfspraul>
first step, I cleanup this project
<wpwrak>
oh, for bragging about our tools to the press. there's also "solidify" in the CAE department ;-)
<wpwrak>
meanwhile, fedex accomplished the miracle of getting my parcel across a third of the planet in only about 24 hours. now it's custom's turn to perform the anti-miracle of sitting on it for a day while doing absolutely nothing ...
<zear>
wpwrak, don't worry, it was weeks of customs in my case
<wpwrak>
zear: lovely customs you have :) btw, my shipment if from digi-key, not a ben
<wpwrak>
iS
<zear>
wpwrak, in my case it was missing invoices, requirement of me signing tons of papers and finally parcel inspection
<wpwrak>
zear: btw2, regarding HoMM2: i wonder, has the artwork also been liberated ?
<zear>
nothing has been
<zear>
it is an engine recreation project
<wpwrak>
zear: (no invoice) oh dear. that's nasty :-(
<zear>
HoMM2 authors didn't do a thing
<wpwrak>
ah, pity
<zear>
so no, their artwork and the rest of the datafiles still aren't libre
<wpwrak>
do the authors even sill exist ? ;-)
<zear>
who knows ;)
<zear>
there were some HoMM series releases recently
<zear>
from what i have heard
<zear>
but it weren't the classic heroes games anymore
<wpwrak>
i have/had HoMM3 for linux. loved the skills system.
<zear>
it's just free software people doing good job recreating the engine, also wrote a new gui for 320x240 devices
<zear>
ah yeah, HoMM3 had an official linux port, am i right?
<wpwrak>
320x240 must have been tricky
<wpwrak>
(official port) yes, by loki
<zear>
it works great with the current gui
<zear>
you should check it out
<wpwrak>
don't tell me this if you want to see anything coming from me in the next months :)
<zear>
looks like the battle screen is scaled down, but the rest of the screens are re-done to fit 320x240
<zear>
the castle screen is totally different (no nice landscape with your buildings), but it still does it's job
<zear>
hehe
<wpwrak>
(re-done for 320x240) sounds great
<wpwrak>
do you know kohan ? also has an official linux port
<zear>
i remember the ads of it in pc magazines
<zear>
i might have a demo on one of my old cds
<zear>
never played it though
<wpwrak>
it's very nice. "real-time" (you can pause, though), 3D-ish view, gridless movements
<zear>
so it's an rts?
<wpwrak>
how, add the skills system of HoMM2 and the complexity of the later members of the rogue family, and you have something that will absorb several generations of geeks :)
<wpwrak>
it's kinda rts. no turns.
<zear>
rogue family as in roguelikes?
<zear>
sounds cool though
<wpwrak>
(rogue family) yup. moria, omega, nethack, the angbands, etc.
<zear>
i prefer turn based strategies though, with the exception of starcraft which i loved back when it was out
<zear>
i'd play anything that reasembles rougelikes in any way ;)
<zear>
which reminds me, there's nethack gui port to the jlime, plays really good
<zear>
i wish one could adopt other roguelikes to this gui
<wpwrak>
strtegy is more precise in turn-based. but kohan is pretty nice. i think with a stronger skill system, you'd have more strategy as well.
<zear>
omega would be hard to port though, i checked the code and it's very binded to the standard size terminal (80ish lines), and makes a big use of carriage return
<wpwrak>
omega may be a bit old, too. how about the later angbands ?
<zear>
i think they already offer sdl interface
<zear>
don't know if one with graphical tilesets
<zear>
but i'm sure i have seen sdl flags during the compilation of zangband
<zear>
that still doesn't mean it could fit in 320x240 though
<zear>
there's powder, which i aready ported
<zear>
but it's more chess than a real roguelike
<wpwrak>
yeah. i also often find the graphical variants of roguelikes not very convincing. hard to see the details.
<wpwrak>
and the graphics often aren't very good.
<zear>
wpwrak, the nethack-newt from jlime is quite cool though
<zear>
by pressing tab you can switch between graphical tileset and classic ascii-like view
<wpwrak>
ah, nice :)
<zear>
it also has some options for different ascii variants
<zear>
the standard ASCII, one from IBM pcs (with smiley faces instead of @) and something else i think
<zear>
some of the text strings don't fit on the screen (the longer fortune cake texts, or item names like "+1 blessed ring of fire resistance"
<zear>
but besides that everything fits perfectly in 320x240
<wpwrak>
you could abreviate item descriptions. e.g., FR. (like some roguelikes have FT weapons)
<wpwrak>
(fire tongue)
<zear>
or they could simply break the lines on the end of the screen ;)
<wpwrak>
or connect another ben for the remaining 320 pixels ;-)
<zear>
:D
<zear>
unfortunatelly don't have a second one ;)
<wpwrak>
it's only a few weeks of customs ;-)
<zear>
:D
<zear>
i don't think i need a second paperweight ;)
<wpwrak>
why - did the first one break ?
<zear>
no, simply don't have much use of it (have a dingoo for playing games, jornada 720 for typing stuff on)
<zear>
if it had wireless or usb host on the other hand..
<wpwrak>
when my parcel makes it out of customs, i'll have some more fun with wireless :)
<rafa>
I need to tell you, that, unfortunately, tuxbrain will not be online for the next two months.. sorry
<rafa>
the resaons are quite complex.. but I will try to say those in short : B_Lizzard and zear showed tuxbrain fheroes2 :D
<tuxbrain>
hehehe, the fact is the lack of save/load of HOMM2 make it unplayable for me.... this or as you said I dissapear for that time
<B_Lizzard>
Oh, does it work?
<wpwrak>
rafa: let's hope victor makes him eat from time to time :)
<zear>
tuxbrain, the game supports save/load just fine
<B_Lizzard>
zear made it compile with a lot of useless stuff, libsdl-net and whatnot
<tuxbrain>
nah I just need cafeine directly on the vein
<rafa>
wpwrak: I asked him if victor likes games, so he should not show him that game.. but it was too late to ask.. tuxbrain was playing
<rafa>
B_Lizzard: yes, we tested your package on muffinman beta3 using demo data files.
<B_Lizzard>
Next up, Settlers II
<wpwrak>
tuxbrain: you know the penguin peppermints ? :)
<zear>
B_Lizzard, you wish there was a 320x240 of settlers II
<tuxbrain>
yep no able to import to spain :(
<B_Lizzard>
There will be when I'm done :D
<zear>
wpwrak, i know it only from hte screenshot. A really crappy use of gimp images ;)
<zear>
B_Lizzard, there will be if you start it
<B_Lizzard>
Or warcraft II
<B_Lizzard>
I can do that
<zear>
B_Lizzard, wargus can't work on 320x240
<B_Lizzard>
I had warcraft II working on the 6xx
<zear>
better start working on the new 320x240 gui
<B_Lizzard>
I even had Baldur's Gate on the 6xx
<zear>
meh, we've got that with a quite fast scaler on the dingoo already
<zear>
it requires a huge memory card to fit all that 6 cd of BG ;)
<B_Lizzard>
:D
<B_Lizzard>
Well, not huge
<rafa>
B_Lizzard: zear : when will you be older and serious??!.. the whole day talking about games
<B_Lizzard>
I can find a 16GB MicroSD for 30$
<zear>
B_Lizzard, i though you greek guys were poor now
<wpwrak>
tuxbrain: (no import) oh, how come ? i used to mail-order them to switzerland
<rafa>
zear: yes they are
<B_Lizzard>
I'm not Greek
<rafa>
red greek
<zear>
B_Lizzard, you're 1/2 greek and 1/4 argentinian
<rafa>
HAHA TRUE!
<zear>
that makes you only 1/4 english
<rafa>
spanish?
<zear>
B_Lizzard's spanish now?
<B_Lizzard>
My mother was born in Spain so that must count
<rafa>
wpwrak: B_Lizzard likes dulce de leche.. you should
<zear>
B_Lizzard, you have ancestors from every nation :P
<rafa>
zear: maybe he is your cousin
<zear>
oh no, must be why i have terrible memories from my childhood :|
<wpwrak>
rafa: that proves his argentine descent then. i think the dulce de leche gene doesn't exist anywhere else :)
<B_Lizzard>
Hah
<wpwrak>
probably the result of some evil experiments escaped nazi scientists did to the unsuspecting population here ...
<B_Lizzard>
This channel is now Godwin'd
<wpwrak>
well, or maybe it was martians. at least it can't be natural
<rafa>
wpwrak: yes.. dulce de leche virus.. and just argentine gene got problems
<wpwrak>
(godwin) took long enough :)
<zear>
B_Lizzard, godwin's law does not apply for discussions about your descent :P
<wpwrak>
maybe it's the result of a chemical warfare project gone awry. maybe the plan as to drop it on the british in the falkland/malvinas way and it somehow got misdirected and ended up in the food chain
<wpwrak>
i.e., if the part was always good for lower voltages but they weren't confident about it, didn't know, or had simply mis-printed it in the spec
<wpwrak>
a part name that begins with "TSOP" is quite clever ;-))
<wpwrak>
no indication of the a different silicon revision. maybe ask them if the old parts also conform to these specs and if not, how they suggest to tell them apart ?
<wpwrak>
of course, if they really messed up and it's impossible, then you'd just have to use the worst-case characteristics. or just avoid the part :)
<wpwrak>
(this is not exactly an exotic type of device)
<wolfspraul>
well let's see
<wolfspraul>
on the first run of 6 Milkymist One, that part was failing and we didn't know why
<wolfspraul>
now we have a clue, will do some tests (we kept all parts), and soon we know more
<wolfspraul>
I'm just presenting you this as a real world example, something a bom system should be flexible enough to handle, all the way from annotated kicad files to the shopping lists at the end
<wolfspraul>
well, if not it's also not the end of the world, but it means that additional manual information needs to be passed around boom
<wolfspraul>
I asked on the list whether anything is printed on the module to tell the two versions apart
<wolfspraul>
but somehow looking at the picture in the datasheet I doubt it
<wolfspraul>
we'll see
<wolfspraul>
if you think about it more generally, you realize that there is an almost infinite list of 'tech specs' a part can be associated with.
<wolfspraul>
and the decision whether to communicate a change and in which way is always a manual decision. and in some cases the decision of the manufacturer may not be compatible with the assumptions your design makes.
<wpwrak>
sure. there can always be parts you can't use just because the manufacturer is unreliable
<wolfspraul>
that's an extreme solution of the problem and may just not be the best/smartest
<wpwrak>
well, if you can educate the manufacturer ... ;-)
<wolfspraul>
it is possible that all participants (manufacturer, distributor, user of the component) know well what they have or need, but there may be a communication breakdown
<wolfspraul>
for this part, for example, I think once we understand this whole thing, we will have no problem sourcing the 'new' version of the module
<wpwrak>
sure. if it's a simple mistake there can be a chance to fix it.
<wolfspraul>
it's just an additional piece of information we need to pass along with the part, when we order, etc.
<wolfspraul>
no. when we order from anybody, we need to pass the following information along:
<wolfspraul>
"unless you know this part is from the newer 2.7V series, we don't want it. we will return it."
<wolfspraul>
this will work beautifully, they will either reject this part because they don't know about this case or which part they have, or because they know they have the wrong one, or they will know they have the right one and sell to you
<wolfspraul>
remember it is in everybody's interest to keep business smooth, no returns, no trouble, etc.
<wolfspraul>
a small mistake in sourcing can cost a lot and ruin the entire relationship with that sourcing partner
<wpwrak>
or they'll just be optimistic and sell it anyway. after all, it's prepaid ;-)
<wolfspraul>
so it is quite common in such cases for them to reject certain parts, because they believe they cannot really get you exactly what you need
<wolfspraul>
no no
<wolfspraul>
wrong
<wolfspraul>
it's the other way round
<wolfspraul>
they rather reject a part too much than one too little
<wolfspraul>
everybody knows what a bad part in a real run means
<wolfspraul>
and what consequences it has for the sourcing people
<wpwrak>
are those pary brokers that careful ? for me, they look like a totally anonymous mass. minimum visible infrastructure, etc.
<wpwrak>
parT
<wolfspraul>
in the current boom, I wouldn't know how to pass such an additional message through
<wolfspraul>
there are .dsc files but they are for the inventory
<wolfspraul>
will prettyorder mix any free-form text field from the bom into the order?
<wpwrak>
in boom, you can identify parts in two ways: by characteristics or by part number
<wpwrak>
prettyorder will just the the description you give it
<wolfspraul>
yes but let's say I want to pass a field from the bom all the way into the order
<wolfspraul>
I think in such a case that's the only way to not have the message get lost
<wpwrak>
you'd have to express this in the part number
<wpwrak>
if you've obtained an assurance from a distributor, you could define that one as an equivalence
<wpwrak>
e.g., call it QIHW TSOP4830-2.7V
<wolfspraul>
and then an equivalence?
<wolfspraul>
hmm
<wpwrak>
then if you find a shop that guarantees you that all the TSOP4840 you order from them are good, then you define SHOP THEIR_NUMBER as an equivalence to your part
<wolfspraul>
where would the text fields prettyorder uses come from?
<wpwrak>
what you can't do is define yours as equivalent to Vishay's, becuase they're not
<wpwrak>
you could have your own .dsc file that describes the part
<wolfspraul>
aren't the .dsc files meant to go with distributor inventories?
<wolfspraul>
yes but I'm not ordering from myself
<wpwrak>
they can come from anywhere :)
<wolfspraul>
so at some point my order will just list some distributor namespace + part number
<wolfspraul>
and at that moment, in that line, the 2.7V info is lost
<wolfspraul>
of course, it was there before
<wolfspraul>
that's why that namespace + part number was chosen
<wolfspraul>
but maybe it would help to pass a free-form text field in here?
<wpwrak>
the 2.7 V info isn't anything special to lose
<wpwrak>
there's a lot of information you lose in all the processes
<wpwrak>
e.g., digi-key's description doens't include the temperature coefficient
<wpwrak>
their part numbers are often just sequential numbers. no way to translate them to any specific characteristics. you can't even tell what sort of thing a, say, 311-2783-ND it is.
<wpwrak>
hah, that one doesn't exist :)
<wpwrak>
i was trying to guess a capacitor or resistor number :)
<wolfspraul>
I can imagine digi-key wants to retain the flexibility to continue to use old or issue new part numbers at a time of their choosing.
<wpwrak>
digi-key's part numbers don't necessarily have anything to do with vendor part numbers. i suppose there may be cases where you have two or more digi-key PNs for the same vendor PN.
<wolfspraul>
yes makes sense to me
<wpwrak>
digi-key's vendor PNs are also a little unreliable. they often put the order code they used, which may not a published code.
<wolfspraul>
like I said, that's probably one important way they believe they create value for their customers
<wpwrak>
i think it's just a basic way of maintaining sanity :) and keeping short PNs
<wpwrak>
my approach is to just take whatever the respective source throws at me and prefix it with a name space. but that can lead to pretty long names.
<wpwrak>
of course, i don't need to print bar codes for them either ;-)
<wolfspraul>
and you are not trying to create certain guaranteed 'service levels' for your customer who buy from you
<wolfspraul>
once you would start reselling I believe over time you would accumulate a lot of reasons why you want to make your own numbers
<wpwrak>
well, i'm agnostic to the 'service level" you want to build on top of everthing :)
<wpwrak>
there's been a discussion about having your own part numbers on kicad-devel
<wpwrak>
it's not a stupid idea
<wolfspraul>
isn't that what you are doing with the way boom uses custom fields already?
<wpwrak>
for BOOM itself, it doesn't matter. it can handle any degree of separation. if each engineer has a PN scheme for every day of the week, the EE department has its own set, sourcing another, etc., BOOM will handle it as long as you provide a suitable rosetta stone
<wpwrak>
you mean the user-defined fields in schematics ?
<wolfspraul>
yes, and the quite flexible processing of all fields in general
<wolfspraul>
reference, value, footprint, user fields
<wolfspraul>
essentially this makes it possible to put any string anywhere and have it picked up by boom
<wpwrak>
yes, you could do fairly crazy things. there are a few limitations, though
<wpwrak>
for example, you can't currently do a part number search on anything but the value field
<kyak>
xiangfu_: hi, are you there?
<xiangfu_>
kyak: ys
<wpwrak>
what you can do is modify the value field for a the 2nd try. (there's a first try before substitutions and a 2nd try after)
<xiangfu_>
yes
<kyak>
xiangfu_: i saw you added some packages to config-full
<wpwrak>
wolfspraul: it works as follows: first, bom2part tries to find a part whose part-number matches exactly what you put into the value field. (the namespace is ignored in this case. this is admittedly a bit dangerous.)
<kyak>
i have some concerns.. for example, ben-cyrillic. It enables users to switch keyboard layout to cyrillic by pressing the "Qi" button
<xiangfu_>
kyak: yes. but I am not test yet. my clean build will start four hours later. then I can check the result tomorrow.
<wpwrak>
wolfspraul: if it finds something, it uses that part (and its equivalences). no further questions asked.
<xiangfu_>
kyak: I am not try ben-cyrillic yet. I just add those new packages today.
<wpwrak>
wolfspraul: if not, it processes the substitutions (.sub). then it tries a lookup with the value field after substitutions (which may or may not have altered the value)
<wpwrak>
wolfspraul: if it still doesn't have anything, it looks for a characteristics match
<kyak>
xiangfu_: my understanding is, if they are included in config-full, they will be included in image by default on the next build...?
<xiangfu_>
kyak: yes.
<kyak>
so that's the point, ben-cyrillic should not be in default image
<wolfspraul>
in the top left there is a component with 6 pins
<wolfspraul>
the component itself is missing
<wpwrak>
yup
<wolfspraul>
the second from top pin on the left side ends in a symbol
<wolfspraul>
is this GND?
<wpwrak>
yes, that's GND
<wolfspraul>
I am now trying to cleanup the project, and in my current local version, that symbol won't show, it just ends in ??
<wolfspraul>
which library is this coming from?
<wpwrak>
the GND symbol doesn't show ?
<wolfspraul>
yes
<wolfspraul>
well
<xiangfu_>
kyak: ok. it's a package of some scripts files. seems russian relationships.
<wolfspraul>
I have removed all system libraries, and Yanjun Luo made some modifications into system libs that I now have in the local tree
<xiangfu_>
kyak: ok. I will remove it from config.full_system.
<wolfspraul>
actually, I can just commit it where it is now
<wpwrak>
it comes from "power". maybe if you don't have any of the default libs installed, then you don't have it
<wolfspraul>
but I'm hesitating to stress test the robustness of schhist
<wolfspraul>
:-)
<xiangfu_>
btw. I added the gcc-mips to config.full_system.
<xiangfu_>
kyak: thanks :)
<kyak>
xiangfu_: might want to add ben-cyrillic as module ?
<wpwrak>
afik, fidelio (sp?) has the default libs, so it should show there
<wolfspraul>
hmm
<kyak>
xiangfu_: so it would build as package, but not included in image..
<wolfspraul>
unless I made a mistake in my new KiCad package, I should have it too
<wolfspraul>
but probably a mistake in the project files
<wolfspraul>
I think I just commit
<wolfspraul>
I trust schhist
<wpwrak>
wolfspraul: click on the 4th icon on the right side (circle and arrow)
<wpwrak>
wolfspraul: aka "Place a power port"
<wolfspraul>
yes
<wolfspraul>
and then?
<kyak>
xiangfu_: i also noticed you included ghostscript, but in this way you should better include fbi and fbgs, i liked fbi as image viewer a lot and fbgs is a "poor man pdf renderer" for SDL, using ghostscript and fbi
<wpwrak>
wolfspraul: then click somewhere on the schematic. a dialog opns.click on list all. do you get something ?
<wolfspraul>
it asks me for a name
<wolfspraul>
list all shows 6 libs
<wolfspraul>
exactly the ones that I left
<wolfspraul>
let me try to add the system power lib back in
<wpwrak>
huh ? list all should show parts, no libs
<kyak>
(i think if we were physically in one room right now it would've been noisy: )
<wpwrak>
perhaps you removed "power" then ?
<wolfspraul>
yes sure
<wolfspraul>
I did
<wpwrak>
hah ! :-)
<xiangfu_>
kyak: ok. thanks, working on that.
<wolfspraul>
great
<wolfspraul>
much less ??
<wolfspraul>
only one left
<kyak>
xiangfu_: ok, and the last concern: i'm afraid gcc-mips and ghostscript might not fit together in rootfs :) they occupy around 80 Mb together; i never tested myself with full image as i have some unneeded (for me) packages stripped off
<wolfspraul>
if I have a ?? in eeschema, how can I tell what it is?
<kyak>
xiangfu_: btw, what would happen with reflash_ben.sh if you supply an oversized rootfs?
<wolfspraul>
I tried all sorts of ways to select it or click on it, but only comparing with the .pdf from schhist gave me a clue
<wpwrak>
yeah, i was about to write "click on it". but then i realize dthat it isn't so easy ;-)
<xiangfu_>
kyak: then we have a reason to increase the rootfs partition :). (let's see what is the size of a clean build )
<wpwrak>
ah, double-click it ;-)
<wpwrak>
then it will tell you what it is inthe "chip name" field
<kyak>
xiangfu_: so, that's all, thank you! :)
<xiangfu_>
kyak: the reflash_ben.sh will overwrite the data partition. "usbboot" doesn't check the rootfs size and it don't know the nand partition.
<wpwrak>
not sure if this helps much in this case, though :)
<wpwrak>
well, if you have only one ?? left, then you found a way to get the other missing bits ...
<kyak>
xiangfu_: wow.. i will pay attention to image size from now on.. don't want to loose my data
<xiangfu_>
kyak: I guess if flash the oversized the rootfs file. the rootfs still can build. because when we build the rootfs.ubi we set the ubi max block to 4096.(which is 2G file)Â Â (but it's just a guess not well test)
<wolfspraul>
now I go from there. my faith in schhist is big.
<wolfspraul>
in terms of file locations, I use ben-wpan as my gold standard
<wolfspraul>
in components, we now have some redundancy with KiCad system libs, conn, device, interface. I will try to remove everything that is also in the system libs.
<wolfspraul>
I can't find this U1/93C46 anywhere, maybe it's still missing entirely.
<wpwrak>
wolfspraul: (std comps) i would go the other way. if there's a local override, consider it a candidate for kicad-libs
<wpwrak>
wolfspraul: that is, find out why there's an override. some of the symbols in the base libs aren't very good or consistent
<wpwrak>
wolfspraul: e.g., many of the power labels have text that's very small in relation to the rest. and it even differs among labels.
<wpwrak>
(the text size)
<wpwrak>
argh. local libs that use the same name as system libs. that's evil :)
<wpwrak>
isn't there enough pain and suffering in the world already ? :)
<wolfspraul>
he :-)
<wolfspraul>
maybe I should have renamed them first
<wolfspraul>
but...
<wolfspraul>
schhist hasn't been stress tested enough, so that will surely come next
<wolfspraul>
first I want to delete all symbols from those libs that aren't used in the project at all, and only there because they were added to a system lib
<wpwrak>
i don't think it'll even notice :) it's more something that confuses the user
<wolfspraul>
not much longer
<wolfspraul>
no worries
<wolfspraul>
do you have any idea about this 93C46 ?
<wolfspraul>
maybe I need to ask Yanjun Luo about it
<wolfspraul>
any suggestion for the name of the library I should move this to?
<wolfspraul>
the conn/device/interface stuff
<wpwrak>
93c46.lib ? :)
<wolfspraul>
huh that number is the name of a .lib file?
<wpwrak>
now i see why the "device" lib is called "device". oh my. a copy of the base lib. some people have no shame :)
<wpwrak>
(number) i would just use one .lib file for each symbol
<wpwrak>
that's what will probably happen in kicad anyway. those "libs" as aggregates make no sense at all
<wolfspraul>
ok, I first need to distill those 3 libs to the real diff
<wolfspraul>
thanks, that is helpful to know
<wpwrak>
well, unless you port it to cp/m for ms-dos 1.0, which don't have subdirectories :)
<wpwrak>
s/for/or/
<wpwrak>
(probably happen) at least the people doing the file format changes talked about it. i very much hope they don't change their mind :)
<wpwrak>
already from a version control/joint editing point of view, those aggregates are evil
<wpwrak>
let's say you and someone else each add one unrelated component to device.lib
<wpwrak>
now you get a merge conflict because the two of you changed the same file. and if you're unlucky, it's a conflict that can't be auto-resolved.
<wpwrak>
bonus points if there's more than one component on each side that got added. now the conflict will be a pretty deep mess to edit :)
<wpwrak>
you'll have the same on the modules side, particularly if you use fped. fped won't even pretent to help you put more than one footprint in a single .mod file :)
<wpwrak>
well, that's actually not true
<wpwrak>
it does, when one .fpd generates multiple footprints.
<wpwrak>
something to fix in the future :)
<wolfspraul>
wpwrak: beautiful! schhist picked it up just fine! :-)
<wpwrak>
as it should :)
<wolfspraul>
wpwrak: would you say that symbols defined in a project can override system symbols (same name), or should they not do this?
<wpwrak>
they can (e.g., for R, C, and such, it would make sense). but then better don't include the system lib that has them too.
<wpwrak>
i.e., i don't know if kicad implements any consistent priority for such cases
<wolfspraul>
ok that's clear (and scary)
<kristianpaul>
wolfspraul: about the chat realated to add a uSD card plus adapter to the Ben i tought it because i think seems to be the easiest way to both provide updates to the nanonote and also carry personal data like music, pics, etcs..
<wolfspraul>
sure understood
<wpwrak>
uSD card + adapter = ?
<wolfspraul>
maybe next run, I have to sell those 1000 first :-)
<wolfspraul>
a think a second battery is even higher priority
<wpwrak>
aah, in the box. i see.
<kristianpaul>
wpwrak: yup
<wpwrak>
uSD cards are easy to get almost everywhere :)
<kristianpaul>
wolfspraul: sure when posible
<kristianpaul>
wpwrak: yes
<wpwrak>
and their value drops rapidly. not a good thing to stock. so this would only make sense if sharism could get a better deal than the local retailers. (taking into account overhead, taxes, etc.)
<kristianpaul>
good point
<wpwrak>
s/better/much better/
<wolfspraul>
wpwrak: there is a point
<wolfspraul>
right now we have almost no distributors
<wolfspraul>
so we also have no distributors that complain that they cannot sell totally overpriced accessories
<kristianpaul>
sure
<wolfspraul>
so there is actually an opportunity to really add meaningful accessories
<kristianpaul>
well maybe a wikipage that suguesst the uSD
<wolfspraul>
to the Ben, also Milkymist or other products in the future
<kristianpaul>
k
<wolfspraul>
and I can source these things at really good prices.
<wolfspraul>
like I said, I would have _almost_ included a second battery in the second batch
<wolfspraul>
only the delay in production run made me not to
<wolfspraul>
for the microSD card, even USB reader, why not
<wolfspraul>
although one problem is that our current box is already very full
<wolfspraul>
the second battery would barely fit
<wolfspraul>
microSD is small, maybe a very small USB reader...
<wolfspraul>
but the truth is also: if I one day have more active distributors somewhere, and they ask me to remove accessories to help them, I would do it
<wpwrak>
hmm, not having .dcms can give you warnings
<wpwrak>
lekernel: seems that you'll have to ask vishay for clarification whether these parts are actually different and how to tell them apart ...
<wpwrak>
it's a little odd that they don't mention the marking in the data sheet. they do that for other components.
<lekernel>
indeed
<wpwrak>
well, i've already noticed that they're not the most thorough company when it comes to documentation. e.g., the si1040x has a pin cleverly marked "ON/OFF" :)
<wpwrak>
(without other attributes, such as an overline anywhere, or such)
<wpwrak>
there are more goodies there :) e.g., a pin names "R1, C1" is quite a rarity, too.
<lekernel>
normally you want the fastest switch times for the power mosfet so it doesn't dissipate when switching
<lekernel>
not a RC charge
<wpwrak>
and then, inthe "typical application circuit", there's one pin that has a dual identity
<lekernel>
wpwrak, yeah, what is C1 for in that typical application circuit?
<wpwrak>
could it be slew control ?
<lekernel>
sounds like it, but it makes the power mosfet dissipate
<wpwrak>
i guess there's no much of a choice if you need slew control ?
<wpwrak>
yeah, says "adjustable slew-rate"
<lekernel>
well it's a simple and stupid solution
<lekernel>
depending on the application it can be good
<wpwrak>
i think "simple" is the idea here :)
<wpwrak>
for most things, you probably don't care. typical application would be a power switch. so you turn the whatever on and then it stays on for a good while.
<wolfspraul>
wpwrak: hmm, now it gets harder. after some cleanup I ran an ERC test, got 5 errors. unconnected pins...
<wolfspraul>
need to see whether that looked better before
<wpwrak>
;-)
<wolfspraul>
then I tried to generate a netlist, but it's quite different from the one that is in git
<wolfspraul>
now, the one that is in git may have been generated with an older version of the schematics anyway, so comparing with that may be useless. maybe the .net in git should just be deleted.
<wolfspraul>
but I'm not sure
<wpwrak>
the netlist doesn't affect the ERC. only the DRC.
<wpwrak>
ah ... overlooked one message
<wpwrak>
(netlist very different) hard to tell. these files generally don't diff nicely.
<wolfspraul>
one of my cleanup steps triggered a schhist diff
<wpwrak>
there are also a lot of ad hoc names that get assigned. not sure where they come from. could be random() ;-)
<wolfspraul>
two lines are gone, I think that looked better before. so maybe another system lib improvement I accidentally reverted.
<wolfspraul>
can you take a look at the schhist output?
<qi-bot>
[commit] Werner Almesberger: c2usb/c2usb.sch: mark design as suspended, due to FT232 being unfit for project http://qi-hw.com/p/f32xbase/07e9667
<wolfspraul>
but it's a lot of text, and sometimes hard to maintain
<wolfspraul>
the cc logo is nice of course...
<wpwrak>
you can auto-export the title just to the other sheets
<wpwrak>
i keep the text block relatively simple. but you can probably do quite fancy things. project url could be nice.
<wolfspraul>
in Milkymist we have, on each sheet:
<wolfspraul>
1. sheet title
<wolfspraul>
2. calendar date
<wolfspraul>
3. feedback email address
<wolfspraul>
4. copyright (author)
<wolfspraul>
5. license
<wolfspraul>
6. cc logo
<wolfspraul>
so it sounds like you think some text on each sheet is ok
<wpwrak>
kicad does take care of the date for you, for better or worse. you get to pick your own version, though
<wolfspraul>
well I would just leave the version empty, date is better
<wolfspraul>
seems you don't have strong preferences one way or the other
<wpwrak>
you may need both. the version is where you tell when to increment
<wpwrak>
instead of mail and license, i'd just use a url
<wolfspraul>
it's probably also better if you write documentation, and people want to know whether the file they are looking at matches the documentation they are reading
<wolfspraul>
also no copyright (author)
<wolfspraul>
just "project homepage"?
<wpwrak>
but i have no strong opinion on the amount of text, as long as it doesn't get in the way of the drawing ...
<wolfspraul>
oh sure same here, totally relaxed. I just try to ping your brain whether you have any strong feelings on it.
<wolfspraul>
I like the cc logo.
<wpwrak>
kicad has an "author" field. so there's your copyright holder (more or less)
<wolfspraul>
gives a quick assurance that it is a free design
<wpwrak>
the logo is nice but probably hard to do ;-)
<wolfspraul>
author could just say 'multiple' (not to discredit anyone, just in case the work is indeed the work of multiple people), then the project URL could give more info
<wolfspraul>
I mean, you can defer a lot of things to the project URL, if you want to
<wpwrak>
(defer) that's the idea :)
<lekernel>
wolfspraul, speaking about that milkymist schematics, could you make again the same pdf from the current (rc2) altium files?
<wpwrak>
if the information you need is small, just put it all there. if not, link to all the non-essential bits. (and hope wolfgang pays his hosting bills ;-)
<lekernel>
so we can publish it
<wpwrak>
ah ... you didn't have f32xbase/hw/ in schist ... so no torture-test then :)
<wolfspraul>
I left your projects the way they were in the demo Makefile, only added the missing projects.
<wolfspraul>
I think I finished the first round of mmone-jtag-serial cleanup, tomorrow I hope to start boomifying a little :-)
<wpwrak>
(boomify) kewl
<wolfspraul>
still need to find one missing symbol in the schematics, and then at some point see whether I can generate gerbers that are at least close to what is committed already. maybe also try to fpedize the one module it has there.
<wpwrak>
(schhist projects) that one was added later. guess you missed it. anyway, now it's in the freezer for a while.
<wpwrak>
having something that goes to real production fped'ized would be great.
<qi-bot>
[commit] Werner Almesberger: c2ben/: initial commit of passive C2 adapter board for the Ben's uSD slot. http://qi-hw.com/p/f32xbase/c1935d5
<wpwrak>
before etching them, i'll cut a few boards. then i can process the whole batch at once.
<arquebus>
hello anyone here?
<wpwrak>
wolfspraul: want a fun read about KiCad and what would be nice to have for state of the art industrial use ? http://www.openss7.org/docs/NOTES.pdf
<wolfspraul>
oh wow
<wolfspraul>
what is that? who will do all this?
<wpwrak>
the author of that encyclopedia seems to be working on some major patches ... that may or may not make it into kicad in the end.
<wolfspraul>
well let's see
<wolfspraul>
:-)
<wolfspraul>
in my experience the people that are making the best plans are not always the ones making the best implementations, and vice versa
<wolfspraul>
but it looks impressive, as you said like an encyclopedia
<kristianpaul>
"Real men lay out their own boards: wimps use autorouters." ..
<wpwrak>
he wrote that his branch is here: lp:~bidulock/kicad/kicad-brian
<kristianpaul>
excuses..
<wpwrak>
kristianpaul: he's not so wrong about that :) well, a semi-automatic can be nice.
<wpwrak>
semi-automatic router
<kristianpaul>
agree
<kristianpaul>
that really stop me to use kicad for routing stuff like fpga
<kristianpaul>
i really admire andres work around xue
<wpwrak>
yeah, xue looks pretty nice
<kristianpaul>
aha !
<kristianpaul>
a missing simbolic link in /etc/rc.d avoid me to mount at startup the data partition and start gmenu2x !
<kristianpaul>
now is this openwrt fault...
<kristianpaul>
how the rc.d is updated
<kristianpaul>
hmm
<kristianpaul>
rafa: there?
<qwebirc32806>
Hello Everybody
<qwebirc32806>
Sooo.....
<kristianpaul>
hello
<qwebirc32806>
I really want to know if anyone... that is anyone.. has witness, or themselves flashed Debian to an SD card for the NN
<qwebirc32806>
there is like 50 ppl in here, and there has to be atleast one
<qwebirc32806>
lol
<qwebirc32806>
Well it atleast feels like 50
<qwebirc32806>
Well I guess that is a negative on being able to flash Debian to an SD card for the Nanonote