<kristianpaul>
Artyom, okay i think i missed that implementation the control register :), btw i noticed you implemnted I/Q separate inputs for namuru, any advantage noticed so far?
<wolfspraul>
wpwrak: thanks for the epiphan link (vga grabber), and yes, I did find them before
<wolfspraul>
problem is that the 300 USD lowest-cost version really only captures 640x480 well (fps), once the resolution goes up fps goes down quite fast
<wolfspraul>
at 1024x768 it could only record 10 fps
<wolfspraul>
so most likely that 300 USD grabber would only be good for a limited timeframe
<wolfspraul>
next, ironically, by the time I have this in China, it will cost at least 500 USD
<wolfspraul>
China is great for anything made here, but has steep import duties for foreign made electronics
<wpwrak>
ah, just like argentina. except that nothing is really made here :)
<wolfspraul>
and it's not like the iphone that is (or was) smuggled into the country in the millions in hand-luggage and thousands of other ways, but for something rare as this, I need to order it and it will be taxed in the official way
<wpwrak>
or find some traveler who can just "hand-carry" it
<wolfspraul>
so for example when I bought my Samsung projector, US price at amazon was something like 420 USD, in Beijing I paid 660 USD
<wolfspraul>
yes totally [hand-carry]
<wolfspraul>
mountains of goods are being imported that way all the time
<wolfspraul>
I have never, not once, seen any customs official at the airport open anything
<wolfspraul>
there's a lot of professional mules who fly all the time for that purpose
<wolfspraul>
and you have no idea what's going on at the HK/Shenzhen border :-)
<wolfspraul>
but... that's not practical for rare devices like this
<wolfspraul>
so, for me it's 500 USD
<wolfspraul>
and the resolution/frame-rate locks me in and will quickly make it obsolete, if we assume progress on m1
<wolfspraul>
it just doesn't feel right
<wolfspraul>
I rather just use a camera for now, and hope that the native m1 resolution goes up, or some form of streaming becomes possible
<wpwrak>
oh, i have some ideas. don't you remember how our HXD8 flashing setup was replicated at the fab ? :)
<wolfspraul>
or maybe the additional support of tv-out or digital video
<wolfspraul>
I rather work on the product itself, than on something like this vga grabber that makes me feel as if we are stuck :-)
<wpwrak>
the problem is that the demo videos don't do the product justice
<wpwrak>
in 1970, they would have been quite alright. but in 2011 ?
<wpwrak>
of course, you could also try to find someone who already has such equipment, and let them do the recording. dunno how difficult that is.
<kristianpaul>
*g*
<wolfspraul>
Sebastien knows someone I think
<wolfspraul>
but whether it will happen is another question, it's very time consuming
<wolfspraul>
we have a lot of patches, we would want videos with different kinds of music, etc.
<wolfspraul>
I will _not_ move forward on the vga grabber
<wolfspraul>
so that opportunity is up for grabs :-)
<DocScrutinizer>
I thought a lot of gfx cards come with video-in?
<wolfspraul>
composite video-in?
<DocScrutinizer>
err yep
<wolfspraul>
m1 currently does not support composite tv-out
<DocScrutinizer>
what else?
<wolfspraul>
it's possible, we sourced adapter cables already, but it needs changes in the Milkymist SoC to support
<wolfspraul>
so many things are possible in the Milkymist SoC :-)
<wolfspraul>
DocScrutinizer: if you run into anybody at the camp good at Verilog, tell them about Milkymist ;-)
<DocScrutinizer>
I think I wil simply send them to the milkymist talk
<wolfspraul>
first of all tell them there is this exciting project, but more people need to work together, rather than on their own isolated island
<wolfspraul>
it's like the real cold fusion hasn't happened yet :-)
<DocScrutinizer>
isn't this supposed to be part of milkymist talk?
<wolfspraul>
1-man project still, at the core of it (the SoC itself)
<wolfspraul>
I would hope, I don't know
<wolfspraul>
it's a bit of a chicken & egg problem
<wolfspraul>
if a second person would show up that could seriously join forces with Sebastien, that'd be awesome
<wolfspraul>
kristianpaul may be one day, with his GPS baseband :-)
<wolfspraul>
but the wishlist in the SoC is really big
<wolfspraul>
tv-out is one of those
<kristianpaul>
i will just the patch.. after that well :)
<wolfspraul>
and real-time video encoding another (that would solve the entire 'grabbing' discussion for good)
<kristianpaul>
i think mwalle already seems to team up well with Sebastien
<wolfspraul>
yes there are hopeful beginnings, we all try, right?
<kristianpaul>
sure
<DocScrutinizer>
just wonders what weird video format MM is supposed to output then
<wolfspraul>
today it outputs vga
<DocScrutinizer>
hah
<wolfspraul>
support of tv-out and s-video is not hard, but still work (according to Sebastien, he said he would do it together with a bigger cleanup/rewrite of the video subsystem)
<DocScrutinizer>
there's been that TV<->VGA converter, for some xx bucks
<wolfspraul>
support for digital video out the same, more work (we would need to add a digital video out connector first, like dvi)
<DocScrutinizer>
I bet it was made in China
<wolfspraul>
well, yes
<wolfspraul>
but it won't work :-)
<wolfspraul>
because it's not implemented in the SoC
<wolfspraul>
I have the adapter with me, and Sebastien has one
<wolfspraul>
now only the software is missing
<wolfspraul>
'only'
<DocScrutinizer>
sorry I can't follow
<wolfspraul>
you can connect that adapter to Milkymist One, but it won't work
<DocScrutinizer>
how's a CVBA<->VGA converter box dependent on anything inside SoC?
<wolfspraul>
ask Sebastien
<wolfspraul>
it's not working today
<wolfspraul>
if it's easy, send a patch :-)
<DocScrutinizer>
I fsckng have no clue what to patch on a equipment and setup that's supposed to *just work*
<aw_>
0x7d histories: 1. After reflash, D2/D3 is dimly lit. 2. http://downloads.qi-hardware.com/hardware/milkymist_one/production/rc3/test_results/7D-reflash-results 3. couple days later d2/d3 is off after power on. 4. d2/d3/ dimly lit after pressed middle btn. 5. stopped at 'Bitstream length: 1484404' after using shorter BEN cable/reflashed. 6. replaced a new diode and c238 7. Let it fully dried over night 8. Reconfiguration->boot
<aw_>
(D2 ON)->rendering(D3 ON) when power on and pressed middle btn
<kristianpaul>
DocScrutinizer: thats not the case for all in milkymist
<kristianpaul>
but it prety do cool video sinthesis :-)
<aw_>
not bad results that replaced new diode and C238. ;-)
<DocScrutinizer>
is *completely* puzzled
<wolfspraul>
*just work* is for vga
<wolfspraul>
you started to talk about tv-out (composite video), and I told you it's possible in hardware, but needs a software (ic design) update
<kristianpaul>
I think i understand DocScrutinizer, i would spect as well that if something is VGA that chinesse adapter just will work
<DocScrutinizer>
no, I talked about TV-IN, NOT TV-out
<kristianpaul>
ah...
<wolfspraul>
kristianpaul: no way, very few graphics cards support that
<wolfspraul>
you cannot plug this into any vga connector and think it will work: it will not
<kristianpaul>
ah, is not just the cable magic ;)
<kristianpaul>
got it
<wolfspraul>
the cable is just a passive part, I sourced it for 1 USD
<DocScrutinizer>
*you* started to talk about TV-out, while I pointed at a converter box available for some 4A$, that converts *any* VGA to CVBS
<kristianpaul>
Artyom, yes, i'm aware of this behavior for STATUS and NEW_DATA, but i was wondering for example about the internal chx_prn_key_enable, how it is disabled..
<DocScrutinizer>
probably you want the second linked one above then, only one time DA->AD -->HDMI
<kristianpaul>
Artyom, oh, wait a second i think i missundertood initialization after reset and initialization after every clk? (from line 276 in your vhdl gps baseband code)
<DocScrutinizer>
>>
<DocScrutinizer>
wandelt analoges VGA in digitales HDMI High Speed um
<DocScrutinizer>
erreicht Auflösungen bis 1080p, ist HDTV kompatibel, 5GBit/s
<DocScrutinizer>
EUR 35,75
<kristianpaul>
Artyom, wich actually differ a bit from the namuru upstream code i have...
<wolfspraul>
DocScrutinizer: yes but we are trying to record. Maybe the other idea works, first vga -> tv-out, then tv-out -> usb
<wolfspraul>
both are cheap little things that I most likely can find in China for a few USD
<kristianpaul>
cant you record from s-video?
<DocScrutinizer>
err, you got no camera with HDMI input for recording? Or a graCa?
<wolfspraul>
no
<DocScrutinizer>
VDR
<kristianpaul>
Artyom, i think i got it now, you reset enables and new_data_read every clk, but i'm amazed this is not on my namuru verilog code, thats why i was thinking the only way was writing back to some registers..
<wolfspraul>
DocScrutinizer: alright, xiangfu ordered both the vga->tv converter (15 USD), and a tv->usb grabber (18 USD)
<wolfspraul>
let's see whether it works and how good the quality is...
<wolfspraul>
thanks for the link, I got stuck at the expensive epiphany vga grabbers before... didn't think of the path through an active tv-out converter
<DocScrutinizer>
yw
<DocScrutinizer>
I searched a bit for recorders with HDMI-INput but couldn't find any
<DocScrutinizer>
for normal PAL/NTSC TV resolution the quality should be fine though
<DocScrutinizer>
just try to use component or SCART for the analog TV signal connection
<DocScrutinizer>
CVBS is probably the worst quality you can get on analog TV
<DocScrutinizer>
those small hosiden are also good quality signal allegedly, think the name is S-Video
<wolfspraul>
DocScrutinizer: somehow I doubt that this whole conversion path will lead to better quality in the end, but we see
<xiangfu>
DocScrutinizer, ok. those two device have S-video
<wolfspraul>
I just need something practical and cheap that works fast, because as important as having good videos is, it's even more important to advance the state of m1 so that m1 itself can produce higher-resolution output, can stream it, etc.
<wolfspraul>
unfortunately Google (for example) makes big press releases about 'open' WebM, but the openess stops right where the Verilog sources of their encoder start
<wolfspraul>
they probably mean open press release, or open marketing
<DocScrutinizer>
hehe
<wolfspraul>
at least Sebastien managed to extract an actual answer from them (that their sources are not open), he can consider himself privileged (others tried and only got silence back :-))
<DocScrutinizer>
you probably will want to test the VGA->TV converter with a real TV first, to make sure the quality is OKish until there
<wolfspraul>
xiangfu ordered both, 35 USD between them, in a few days we can see the results
<wolfspraul>
:-)
<wolfspraul>
like I said, what we really need is encoding and streaming right on m1 itself, so like I said earlier - keep your eyes open for Verilog hackers :-)
<DocScrutinizer>
all days better than camera shots from a VGA display
<wolfspraul>
maybe there is something simpler than full theora/webm/dirac encoding, something that just streams unencoded over Ethernet?
<wolfspraul>
don't know
<wolfspraul>
the frame grabber should grab it right out of m1 memory :-)
<DocScrutinizer>
umm, don't think so
<wolfspraul>
not through all these external converters...
<DocScrutinizer>
hmmm yeah, sure
<DocScrutinizer>
first see what it yields, maybe you'll be surprised
<wolfspraul>
at 640x480, 16bpp and 24fps, we have about 14 megabytes of data per second
<DocScrutinizer>
honestly I don't get it why MM can't output at 50fps/16kHz rather than 32kHz VGA
<DocScrutinizer>
afaik VGA is TV RGB on double speed
<DocScrutinizer>
so reduce the clock to 1/2 and you should have fine component TV signal
<DocScrutinizer>
just made for SCART
<DocScrutinizer>
ok, you won't get interlace this way
<DocScrutinizer>
and that's also the difference to VGA
<DocScrutinizer>
VGA is basically "progressive" on double Hsync freq, or sth like that
<DocScrutinizer>
yeah, for the interlace you need a little trick to mark each other frame as odd one
<DocScrutinizer>
and your videobuffer scanning scheme is a bit weird
<wolfspraul>
this is all understood, and you are right
<wolfspraul>
it's 'not much' work, but it's work
<DocScrutinizer>
yup
<wolfspraul>
it's one thing to think it through and describe in irc, another to send a functioning Verilog diff
<DocScrutinizer>
Verilog :-)
<wolfspraul>
but of course your thoughts are very appreciated, you hit it exactly
<wolfspraul>
Sebastien said next time he cleans up/restructures the video subsystem, he'll add tv-out with it. that's the last I heard on this.
<DocScrutinizer>
you need somebody to hack interlace tag and videobifer scanning into verilog?
<wolfspraul>
it's not super urgent I agree, just a nice feature
<DocScrutinizer>
should be simple enough
<DocScrutinizer>
I think the odd-frame tag is a "half" line with a long hsync black shoulder
<DocScrutinizer>
so old analog TV would advance vertical sawtooth oscillator by one line, while scanning one line and flyback usually allows vertical to advance 2 lines during that period
<DocScrutinizer>
actually it's the vsync black shoulder that's extended by one half line duration
<DocScrutinizer>
black shoulder for even frame starts and ends half a hsync phase earlier than that for odd frame
<kristianpaul>
(ethernet) cant achieve more than 512Kb/s last time i check
<kristianpaul>
because the lack of dma lekernel said
<wpwrak>
plan B would be the simulation route. if M1 has a fixed frame rate, it would be a matter of processing the audio stream, extracting the control (input to the gfx) information frame by frame from it, then individually rendering the frames, which qemu apparently already can do (render, i mean, not individual frames)
<wolfspraul>
wpwrak: yes, we looked at that too, but it's ... more work :-)
<wolfspraul>
how to feed audio to qemu, let alone camera...
<wolfspraul>
the best route is to make things work right inside m1
<wolfspraul>
and keep investments into workarounds to a minimum, because they all get deducted from m1 investments
<wolfspraul>
of course, I know, we need videos... just saying the emulation route was considered
<wpwrak>
wolfspraul: yeah, the more direct, the better :)
<Artyom>
kristianpaul: (from line 276 in your vhdl gps baseband code) - this was done because of difference in connection FPGA and MCU on my board. The idea of this code is to make pulse that lasts only one-clock-period for these signals. But I don't know whether this is correct implementation.
<Artyom>
kristianpaul: I have added both I/Q-channels because I would like in future to add GLONASS implementation (because it's C/A-code bandwidth is wider).
<Artyom>
kristianpaul: yes, I took namuru source-code from the same place.
<xiangfu>
video-in itself make performance interesting .
<larsc>
xiangfu: hehe :)
<LunaFrizzle>
Hi everyone !
<LunaFrizzle>
Hello xiangfu :)
<xiangfu>
LunaFrizzle, hi
<wpwrak>
xiangfu: oh. took me a while to realize what this screenshot is showing ;-) so this vga->video in converter will also be useful for recording demo videos, right ?
<wpwrak>
xiangfu: if yes, it looks pretty darn good :)
<xiangfu>
wpwrak, yes. but I still not get the usb-Video-grab device.
<wpwrak>
xiangfu: relax ;-)))
<wpwrak>
in any case, what your screenshot shows is the same as what such a device does, i.e., the same type of signal path. just that the destination is the M1 itself, not a PC
<wpwrak>
xiangfu: can you set a background image ("wallpaper") for the screen instead of just black ? something colorful ?
<xiangfu>
ye
<xiangfu>
yes
<wolfspraul>
pretty brave to plug the cable back into m1 itself :-)
<wolfspraul>
xiangfu has no fear
<wpwrak>
wolfspraul: it's a pretty clever idea :)
<wolfspraul>
mental stack overflow
<wpwrak>
wolfspraul: and see how well it seems to work !
<wolfspraul>
I'd be worried...
<wolfspraul>
:-)
<wpwrak>
now, with a colorful background, we could also tell how well color information is preserved
<xiangfu>
wait one second
<wpwrak>
wolfspraul: what's also cool about this setup is that you can directly compare, side by side, the effect of 1-2 times going through the analog loop
<wolfspraul>
indeed
<wolfspraul>
it also has a feeling of time machine to it
<wolfspraul>
and electrons going in circle
<wpwrak>
wolfspraul: i think you already have your first item for the 09-01 news ;-)
<wolfspraul>
and now you want to be able to zoom in...
<wolfspraul>
can we catch up?
<wpwrak>
;-))
<DocScrutinizer>
xiangfu: optical feedback - nice
<wolfspraul>
DocScrutinizer: you see that your input gives a quick echo here... xiangfu already received the adapter you suggested a few hours ago...
<wpwrak>
i'm amazed by the speed of that as well
<wpwrak>
overnight delivery is so yesterday
<DocScrutinizer>
yes, and the feedback setup is a brilliant method to amplify any artifacts introduced by any part of the system
<wolfspraul>
in China it's not that unusual
<wpwrak>
wolfspraul: add "'s empire" :)
<wolfspraul>
keep in mind that most Chinese cities are big, Beijing alone is bigger than most smaller European countries
<wolfspraul>
so as long as you buy within your city, there's always a delivery option that will get stuff to you within 2 hours, if you want to (will cost more, but not much)
<wolfspraul>
then some kid will race through town on his bike to deliver your stuff :-)
<wolfspraul>
if you want to, that is (it will cost more)
<wolfspraul>
so xiangfu was lucky that that one vendor was in Beijing, the other one probably not, so that gets delivered a little slower
<wpwrak>
yeah, and in the US they prefer to have shops in states with a few customers as possible, to evade the sales tax. so there, shipments are generally non-trivial.
<wolfspraul>
if you are in shenzhen, you will have hardly heard about ANY electronic anything needing more than 2-3 hours to reach you, as long as you pay a few USD for a courier
<DocScrutinizer>
wolfspraul: anyway you see there's not much loss on high frequencies (small structures smearing, edges fuzzy), and no change in hue and saturation
<wolfspraul>
well
<wolfspraul>
we are following your advice
<wpwrak>
i like that. this way, home delivery begins to make sense
<wolfspraul>
xiangfu's last round of videos was a step up (camera on tripod)
<DocScrutinizer>
even when running things three times thru the whole AD->-DA process
<wolfspraul>
and if this is yet another step up - great
<wolfspraul>
the 300+ USD vga grabbers looked wrong to me
<wolfspraul>
now we are down to 30 USD
<wolfspraul>
good
<wpwrak>
hehe :)
<wolfspraul>
if this works (we only have one half now), I'll buy a few more sets for people who want to help with tutorials and recordings
<wolfspraul>
wpwrak: yes it's nuts in shenzhen. I tried to do some rough math once and the value of components _IN STOCK_ within a maybe 1 mile radius of only one of the electronic shopping areas must be in the hundreds of millions of USD
<wolfspraul>
and within a range of about 50km, ca. 25% of the world's (!) consumer electronics are being manufactured, by tens of thousands of hungry smt machines hammering away
<wolfspraul>
those 2 numbers give you an idea of the scale, and why you can get pretty much anything delivered to your office in a few hours
<wolfspraul>
and quick and flexbile reaction is so important in any run...
<xiangfu>
the by-pass VGA out, have a little noise. but not much. the screenshot can show that.
<xiangfu>
LunaFrizzle, ok. but I am not SDL expert. I think someone in the mailing list is SDL expert. :) what is your error?
<LunaFrizzle>
xiangfu, first off, to be sure the file have to be compiled on my computer or on the Nanonote (I got the same error anyway, probably a dumb error)
<xiangfu>
this device support PAL and NTSC
<xiangfu>
here is NTSC screenshot: (it show NTSM-MJ) is m1.
<xiangfu>
a little bigger then PAL in m1 [Video input preview]
<xiangfu>
LunaFrizzle, the 'main.c' cannot compiled? that is your compile commands?
<xiangfu>
LunaFrizzle, oh. sorry. you also using the same Makefile in that wiki page?
<wolfspraul>
interesting
<wolfspraul>
ntsc fills out a larger part of the preview window
<wolfspraul>
now we can also easily test pal & ntsc :-)
<wpwrak>
LunaFrizzle: if you don't tell anyone what errors you get, i'm afraid this problem will haunt you for a very long time :)
<LunaFrizzle>
sorry xiangfu, wpwrak I'm first digging in myself to be sure I'm not waisting your time
<wpwrak>
yellow fades quickly but the other colors look quite decent, even in the second iteration.
<DocScrutinizer>
wolfspraul: anyway the VGA->TV converter looks pretty decent, I'd think with a proper digitizer you shall get acceptable quality for any demo videos
<wpwrak>
LunaFrizzle: a request for help with the error message is likely to waste less time than one without ;) you know, it's that sherlock holmes thing :)
<xiangfu>
wolfspraul, it have a little shift at the bottom when NTSC
<xiangfu>
wolfspraul, in the Video-in-Preview.
<wpwrak>
NTSC also seems to "overshoot" at the bottom
<xiangfu>
maybe a bug.
<wpwrak>
probably :)
<wpwrak>
in any case, i think the days of lousy demo videos are counted ;-)
<DocScrutinizer>
isn't it amazing what you can get for some 13$? :-D
<wpwrak>
DocScrutinizer: by the way, what do the pros use if they want to power a little uC circuit from mains ? constraints: voltage 3.3 V, about 50 mA max, long-term average (> 1 min) very low, has a triac to switch mains (it's mainly a lamp control), circuit should be simple and should use as little volume as possible (needs to share the box in which the light switch normally lives with all the cables and stuff)
<wpwrak>
DocScrutinizer: things i've considered: resistive divider, capacitative divider, transformer
<wpwrak>
DocScrutinizer: i can't seem to find transformers that are really small. the smalles one is still some 6 ccm
<wpwrak>
DocScrutinizer: also, since i have the triac, i'm likely to be galvanically coupled to mains anyway (unless i use an opto-isolated one, but these have other undesirable properties)
<DocScrutinizer>
capacitive divider usually
<wpwrak>
DocScrutinizer: great. that was the approach that looked the most promising
<wpwrak>
DocScrutinizer: i wonder if it would be a good idea to toss in a switching DC-DC regulator and convert a relatively high input voltage (maybe 12-24 V) down to the 3.3 V I need, to keep the current on the high-voltage side low ?
<DocScrutinizer>
nah
<DocScrutinizer>
50mA is a lot though
<DocScrutinizer>
those capacitive dividers are feasible for up to maybe 5mA
<DocScrutinizer>
if those 50mA are not spikes you can buffer in a 1000uF, you probably *will* need a trafo
<DocScrutinizer>
or a special solid state converter: there is at least one IC that can (switch?)regulate down from 300V input to 1digit volts output, but I think it also doesn't do 50mA
<DocScrutinizer>
complete SPSU modules should be available as well
<DocScrutinizer>
so not 24V->3V3 but rather 300V->3V3
<DocScrutinizer>
BOM is almost identical ;-D
<wpwrak>
hmm, i haven't found any regulator rated for such high voltages
<wpwrak>
also, how big are those modules ? remember, i just have a few ccm
<wpwrak>
about 1/10th the volume may be be acceptable, but just barely
<stefan_schmidt>
sits at the cccamp and is watching daniel and jan to get the GSm up and running
<wpwrak>
i have about 9 x 5 x 4 cm inside these boxes. and that's shared with all the cables going in and out
<wpwrak>
stefan_schmidt: there's hardly any other thing in life as satisfying as watching other people work, while sitting back, relaxing, and enjoying a cool beer, right ;-)
<stefan_schmidt>
wpwrak: heh
<stefan_schmidt>
wpwrak: actually I help as well
<stefan_schmidt>
wpwrak: But a cold beer would not be that bad
<stefan_schmidt>
wpwrak: Harald still have some small 5l kegs around...
<wpwrak>
just the right quantity to wash down the breakfast
<LunaFrizzle>
ok I got my problem
<LunaFrizzle>
I compiled the toolchain following the wiki's intruction
<LunaFrizzle>
but
<LunaFrizzle>
it doesn't compile the bin in the staging_dir/toolchain-mipsel_gcc-4.3.3+cs_uClibc-0.9.30.1/usr/ folder
<LunaFrizzle>
I don't know what I got wrong
<kyak>
it's in staging_dir/toolchain-mipsel_gcc-4.3.3+cs_uClibc-0.9.30.1/bin
<LunaFrizzle>
there's no such folder either kilae
<LunaFrizzle>
I mean kyak
<LunaFrizzle>
xiangfu is gone but here is my problem
<wpwrak>
kyak: for me, regular makefiles work quite well for cross-compiling basic things with openwrt (not SDL).
<wpwrak>
kyak: if the unfortunate split between gcc and the libraries hierarchy didn't exist, also SDL wouldn't be a problem
<wpwrak>
kyak: (what i did was simply symlink all the binaries to /usr/local/bin. gcc remembers its "true home", so it goes back there to look for things)
<kyak>
wpwrak: for me, openwrt Makefiles work quite well, without further remarks :)
<wpwrak>
yeah, but it'snot so nice if you have to modify your makefiles a lot for openwrt. it's a bit like as if Debian and Red Hat would need different Makefiles ...