<mithro>
felix_: ping? Message me when you are round?
stekern has joined #m-labs
Mon_ has joined #m-labs
Mon_ is now known as Guest19535
Guest19535 has quit [Client Quit]
Mon_1 has joined #m-labs
fengling has quit [Ping timeout: 245 seconds]
<felix_>
mithro: pong
Mon_1 has quit [Quit: This computer has gone to sleep]
Mon_ has joined #m-labs
Mon_ is now known as Guest48154
Guest48154 has quit [Quit: This computer has gone to sleep]
Guest48154 has joined #m-labs
fengling has joined #m-labs
Guest48154 has quit [Quit: This computer has gone to sleep]
rohitksingh has joined #m-labs
rohitksingh has quit [Ping timeout: 264 seconds]
rohitksingh has joined #m-labs
Mon_ has joined #m-labs
Mon_ is now known as Guest89050
Guest89050 has quit [Quit: This computer has gone to sleep]
<sb0>
_florent_, are you using with_identifier in misoc?
<sb0>
I'm trying to cut the too many options that the base soc classes have... the idea is if your the soc differs to much from that base, the user probably should take another one
<GitHub46>
[misoc] sbourdeauducq pushed 3 new commits to new: http://git.io/vlX8h
<_florent_>
sb0: yes I have a specific identifier in one design
<_florent_>
I'm still wondering what will be the best way for me to use new migen/misoc.
<sb0>
_florent_, what's specific about it?
<_florent_>
The idea I have for now is to have my own repository with migen/misoc as submodules that implements the missing parts I need for my designs:
<_florent_>
Verilator simulation, some dataflow modules, custom integration schemes)
<_florent_>
So going that way means that you can simplify things on your side
<_florent_>
yes sorry as packages
<sb0>
verilator simulation is mergeable if architected properly and no longer does things like os.system("openvpn....")
<sb0>
_florent_, what do you need for the identifier which isn't already there?
<whitequark>
verilator support did WHAT?
<sb0>
whitequark, that was for creating a tun/tap interface to connect the simulated ethernet core to an interface on the host
<_florent_>
sb0: sorry I don't remember exactly for the identifier, but you can remove the with_identifier parameter
<sb0>
_florent_, so the current identifier core is ok for you?
<sb0>
we might add some features to it e.g. to track bitstream versions
<_florent_>
IIRC I added a custom revision with specific encoding (not based on git id), that was for my PCIe design
<_florent_>
but looking at it now, I could have used the provided identifier
fengling has quit [Ping timeout: 245 seconds]
<cr1901_modern>
sb0: Do you anticipate having to create more Qt custom widgets?
<cr1901_modern>
I think it would be best for me to create a "widgets" directory under artiq/gui; write the widget itself in PyQt.
<cr1901_modern>
rjo_: In the picture that presumably either you or your colleagues drew here: https://goo.gl/photos/AG7fPLGoACcZ8h4W9, is there any significance to the "A" or "F" on the left side of the image?
fengling has joined #m-labs
fengling has quit [Ping timeout: 245 seconds]
ylamarre has joined #m-labs
Gurty has quit [Read error: Connection reset by peer]
<sb0>
cr1901_modern, pyqt seems like the right choice for writing this, yes. i hope you are not using c++ ...
<sb0>
cr1901_modern, we might have to fork pyqtgraph since it seems rather neglected. if that happens, it would be the right place for that widget
<sb0>
cr1901_modern, in the meantime, put it in its own module in artiq.gui ...
Mon_ has joined #m-labs
Mon_ is now known as Guest79404
Guest79404 has quit [Client Quit]
<sb0>
cr1901_modern, A = fit view to markers, F = fit markers to view
<sb0>
they're small buttons like the existing A in pyqtgraph. feel free to use better letters or icons.
Mon_1 has joined #m-labs
Gurty has joined #m-labs
Mon_1 has quit [Quit: This computer has gone to sleep]
Mon_1 has joined #m-labs
<sb0>
why is git such a clusterfuck with moving submodules?
<sb0>
whitequark, where is $HOME/.m-labs/build_settings.sh coming from?
<sb0>
ah, the .travis scripts.
fengling has joined #m-labs
fengling has quit [Ping timeout: 245 seconds]
<sb0>
whitequark, why are the device files called /bios.bin, /artiq_kc705-nist_qcX-kc705.bit and /nist_qcX/runtime.bin? is that some legacy from artiq_flash.sh?
<sb0>
i'd use /board_nist_qcX/{top.bit, bios.bin, runtime.fbi} instead...
<whitequark>
sb0: I just modified the existing script
rohitksingh has quit [Ping timeout: 250 seconds]
ylamarre has quit [Ping timeout: 265 seconds]
rohitksingh has joined #m-labs
<rjo_>
cr1901_modern: that was me. F is "fit markers to view", A is the inverse of F: "fit view to markers".
<whitequark>
sb0: anyway I will change it to be as you want
rohitksingh has quit [Ping timeout: 240 seconds]
<rjo_>
cr1901_modern: are clear on the details of what this thing is trying to achieve and the potential pitfalls/corner cases?
<rjo_>
A is more like the adjoint of F...
fengling has joined #m-labs
fengling has quit [Ping timeout: 245 seconds]
rohitksingh has joined #m-labs
rohitksingh has quit [Ping timeout: 250 seconds]
rohitksingh has joined #m-labs
rohitksingh has quit [Ping timeout: 240 seconds]
rohitksingh has joined #m-labs
rohitksingh has quit [Client Quit]
<cr1901_modern>
sb0: ACK. Doesn't make sense to write the widget in C++ when artiq by and large doesn't depend on it.
<cr1901_modern>
rjo_: ACK.
<cr1901_modern>
I expect PyQt should expose enough of Qt to me such that I can make the widget. Maybe even without opening Designer :P
fengling has joined #m-labs
<cr1901_modern>
rjo_: I understand what the purpose of the widget is (it complements the spinboxes of artiq/gui/scan.py). Not sure about pitfalls except except the case where "N" is made too large and the tickmarks start becoming indistinguishable.
fengling has quit [Ping timeout: 245 seconds]
rohitksingh has joined #m-labs
rohitksingh has quit [Quit: Leaving.]
<rjo_>
cr1901_modern: a few things come to mind: how would one handle logarithmic scales/steps? what if there is just one point (start=end)? can there be some "smart snapping" to somewhat "rounded" values? etc.
<rjo_>
cr1901_modern: the abstract notion is "intuitively and quickly define/change/browse/verify/visualize a range(start, stop, step)"
<cr1901_modern>
rjo_: Re: start = end: Assuming that's a legal combination, that'll have to be handled as a special case if a person decides to click "fit view to markers"
<cr1901_modern>
One way I can think of handling that is to do what MATLAB does for plots that are zoomed "too far in" (picture's worth 1000 words, so hold on):
<rjo_>
well. one would first start to define what the representation of the information is (step or N), then whether it includes the enpoint, then what to do if the endpoint does is not "hit", then how to represent zero-length and one-length sequences etc.
<rjo_>
please don't look at matlab too much. it will confuse you. the way they do GUIs there is abysmal.
<rjo_>
and then there are the possible "actions" that you can do on the widget: how do I move the entire range (start and end), does it auto-scroll/zoom if i drag a marker beyond the widget...
<cr1901_modern>
I'm writing notes into a notepad about how to implement before I jump into creating anything besides what the widget looks like
<rjo_>
good. i can also recommend scrolling through how gnome/kde/ubuntu mir etc sketch/draft/design/define/refine thei user interfaces.
<rjo_>
and then start > end (or negative N/step) is also a useful case
<cr1901_modern>
Since spinboxes are synchronized to the graph, that should be illegal to do
<cr1901_modern>
i.e. if you try it, it won't honor.
<cr1901_modern>
Unless you ave a legit use case for it
<cr1901_modern>
And step or N? What's the difference? Your drawing only discusses N.
fengling has joined #m-labs
<rjo_>
scanning reverse (large to small) has several legitimate use cases.
<rjo_>
the drawing does not discuss much at all. i just sketched a few things. in most cases people are fine with N, but I can see cases where step is much more helpful. but that would only be the visual represenation. the programmatic representation does not need to follow the visual.
<whitequark>
conda actually fixed the bug I've bumped into
<cr1901_modern>
rjo_: Let me build a nonfunctional widget tonight and get back to you re: step vs N. I still don't understand how it should be visualized, but once I have a base widget to work with, it'll likely make more sense.
ylamarre has joined #m-labs
fengling has joined #m-labs
fengling has quit [Ping timeout: 245 seconds]
ylamarre has quit [Remote host closed the connection]