<Joerg-Neo900>
indeed long time. I hope to contract the layouter today
Kabouik_ has joined #neo900
Kabouik has quit [Read error: Connection reset by peer]
Kabouik_ has quit [Remote host closed the connection]
Kabouik_ has joined #neo900
<Venemo_>
oh
<Venemo_>
the layout is done by someone else?
<Joerg-Neo900>
yes
<Joerg-Neo900>
layout is a very special skill
<Venemo_>
but is it automated or done by hand?
<Venemo_>
I've designed PCBs before, but not of this complexity
<Joerg-Neo900>
for layout there are autorouter programs but they fail epically on anything mildly more complex
<Venemo_>
yeah, I thought so
Pali has joined #neo900
<wpwrak>
Venemo_: hmm, looks as if some error checking changed in gitlib. which version do you use ?
<Venemo_>
wpwrak: latest release from the fedora 25 repo
<wpwrak>
do you know version number that is ?
<wpwrak>
insert "what" :)
<Venemo_>
wpwrak: it's 0.24.6
<Venemo_>
is that too old or too new? :)
<wpwrak>
hmm, i have 24.1. let's see if there's a documented change
<wpwrak>
err, i mean libgit2, not gitlib. gitlib would be something else.
freemangordon has joined #neo900
<Venemo_>
yes
<Venemo_>
wpwrak: I misunderstood it for libgit2
<Venemo_>
so that's it :)
<wpwrak>
yes :) with gitlib, the version number would be quite different
<Venemo_>
is it possible that eeshow simply searches for the .git location in the wrong place?
<Venemo_>
at least, the output suggests that it tries to parse the .pro file as if it were a .git
<wpwrak>
hmm, they tightened some checks in may. might be that.
<wpwrak>
what i do is that i throw the path of a file to git_repository_open_ext and rely on it to find the repo for me. seems that it no longer does that.
<Joerg-Neo900>
>>The `.git` file at '/home/Timur/Projects/smog1/smog1-obc/obc-hw/smog1-obc.pro' is malformed<< actually sounds slightly odd
<Joerg-Neo900>
there's no such thing
<Joerg-Neo900>
wpwrak: use dirname?
<Joerg-Neo900>
aah, I think I understand
<wpwrak>
yeah, i'll have to use the dirname
<Joerg-Neo900>
is the error message from libgit2?
<Joerg-Neo900>
:nod:, needs dirname(): >>Path to open as git repository. If the flags permit "searching", then this can be a path to a subdirectory inside the working directory of the repository.<<
<Joerg-Neo900>
wpwrak: when venemo confirms, could you please change window title from "eeshow (rev XXXXX)" to "eeshow (stable-rev XXXXX)" and to a commit tagged "stable"?
<Joerg-Neo900>
s/and to/and do/
<Joerg-Neo900>
distro maintainers need *some* sort of idication which app version they may add to their next release
<Joerg-Neo900>
actually the window title editing isn't even needed, only the commit with tag
<wpwrak>
you know i don't do "stable" ;-) if i make changes that i think have a high risk of upsetting things, i do them in a branch. so "master" is always as good as it gets
<Joerg-Neo900>
nobody knows that
<Joerg-Neo900>
and tight now that statement is not necessarily correct
<Joerg-Neo900>
right now*
<wpwrak>
well, unless the fix is confirmed, master is still as good as it gets :)
<wpwrak>
just has one known bug i didn't know about until a few minutes ago
<Joerg-Neo900>
please understand that distro maintainers need a minimal amount of hints regarding which app versions they may add to their repos/releases
<Joerg-Neo900>
even if your approach was absolutely fine for them, they simply don't know about it
Oksanaa has joined #neo900
Oksana has quit [Ping timeout: 255 seconds]
<Joerg-Neo900>
to get a app version into any arbitrary distro release, there either needs to be a ('anonymous') pull request to distro maintainers pointing to a app version tagged "stable", or the pull request has to come from developer himself then the version he asks to pull is implicitly the stable one
<Joerg-Neo900>
you could do a pull request "to MASTER, which is always considered stable" and hope for the mainteiner adding a note about that unusual policy so they use master next time without needing a pull request
<Joerg-Neo900>
but you can't expect distro maintainers to guess about that policy, it's nowhere announced for that project
Venemo_ has joined #neo900
<Joerg-Neo900>
you may add the policy verbatim to the readme
<Joerg-Neo900>
"MASTER always stable"
<Joerg-Neo900>
so anybody can point to that when doing a pull request on our behalf
<Joerg-Neo900>
so maybe let me change my request above: please edit README to add a few words about your policy that considers MASTER always stable, and make a commit with the comment "MASTER always stable"
<Joerg-Neo900>
that should be fine for everybody, I think
Venemo_j has quit [Ping timeout: 258 seconds]
Venemo_j has joined #neo900
Venemo_ has quit [Ping timeout: 260 seconds]
<Joerg-Neo900>
I could do that commit, editing readme by adding a quote of your line above ([2017-06-14 Wed 18:21:14] <wpwrak> you know i don't do "stable" ...) but that seems a tad silly. Better you do that
<wpwrak>
ah, it goes past it. good. let's see where that happens then ...
<Venemo>
same as before
<Venemo>
it basically just opens the latest revision, but I don't see any more versions
<wpwrak>
does this show the history ? (one commit per line) eetest -H /home/Timur/Projects/smog1/smog1-obc/obc-hw/smog1-obc.pro
<wpwrak>
eetest is built as part of eeshow, but not installed by "make install". so you have to pick it from the build dir
<Venemo>
no
<Venemo>
it just says the same thing
<Venemo>
this: /home/Timur/Projects/smog1/smog1-obc/obc-hw/smog1-obc.pro: The `.git` file at '/home/Timur/Projects/smog1/smog1-obc/obc-hw/smog1-obc.pro' is malformed
<wpwrak>
okay, that narrows down the problem
<wpwrak>
ah .. does the file /home/Timur/Projects/smog1/smog1-obc/obc-hw/smog1-obc.pro exist ?
<wpwrak>
(file exists) eeshow can also sometimes work on file that don't exist (e.g., if they existed only at a different point in history). that's why i asked.
<Venemo>
oh, now it's doing something
<Venemo>
I see a progress bar
<wpwrak>
good ! :)
<Venemo>
awesome, I think it's working now
<wpwrak>
great. thanks for the bug report !
<Venemo>
wpwrak: you're welcome. what was the issue after all?
<Venemo>
that one liner?
<wpwrak>
yup. i did already try to trim the path ... but due to that typo, didn't
<Venemo>
ahh
<Venemo>
okay, now
<Venemo>
I'm not quite sure what I'm looking at
<wpwrak>
your schematics ? :)
<Venemo>
I guess the red lines mean deletion and green addition
<wpwrak>
yup, when viewing differences it's like that
<Joerg-Neo900>
try "?"
<Joerg-Neo900>
sorry "h"
<Joerg-Neo900>
and D N O
<Joerg-Neo900>
and Tab
<Venemo>
cool
<Venemo>
what does yellow mean?
<Joerg-Neo900>
modified
<Joerg-Neo900>
in index sheets
<Joerg-Neo900>
so you find differences there
<Venemo>
ah indeed, the reference changed
<Venemo>
didn't notice that at first glance
<Venemo>
hm
<Venemo>
brilliant tool you guys made
cc___ has joined #neo900
<Venemo>
wpwrak: thanks for your help
<Joerg-Neo900>
in yes, wpwrak is a genius
<wpwrak>
you're welcome :)
<Joerg-Neo900>
one thing I'm missing is to have git hash in file parameter, to have -N versions back from that particular (instead MASTER) version of the project
<Joerg-Neo900>
like in eeshow -N 10 61ba4e4bfee2f54db1c7636137394e98e735629b:neo900.pro
<Venemo>
what does the pink color mean? does it just simply highlight the modified area, or does it mean something more?
<Joerg-Neo900>
assume 61ba4e4bfee2f54db1c7636137394e98e735629b is a hundred commits ago
<Joerg-Neo900>
right, just highlight
<Joerg-Neo900>
though it may highlight other things too
<Venemo>
I also see that it tries to detect which commits contain actual schematic changes and which not
<Venemo>
but it does show some commits which certainly didn't
<Joerg-Neo900>
it shows all commits afaik
<Venemo>
no, it doesn't
<Venemo>
we have the hw and sw in the same repo
<Joerg-Neo900>
well "3 commits without changes"
<Venemo>
eeshow filters out commits which didn't have changes to the hw, and just shows "N commits without changes"
<Venemo>
but it has some false positives
<Joerg-Neo900>
I doubt that :-)
<Joerg-Neo900>
maybe the 'changes' had no visible effect
<Joerg-Neo900>
kicad is terrible with that
<Venemo>
there has literally been no changes to the hw for quite a while now
<Joerg-Neo900>
I guess for detecting commits with changes it only checks which have a new version of relevant files, even when those files only differ in a whitepsace or whatever
<Joerg-Neo900>
or for example one of the (invisible) timestamps kicad uses in everything
<Venemo>
according to git itself there had been no changes to that directory since last september
<Joerg-Neo900>
open in kicad, do "no changes", save, commit. -> new version in git
<Joerg-Neo900>
hmmm
<Venemo>
yes I know, but as I said the files have been unchanged in git for a long time
<Joerg-Neo900>
werner might be interested in a test case for that
<Venemo>
unfortunately I can't disclose this project (at least not yet)
<Joerg-Neo900>
np, you might provide the -vvv logs or whatever
<Venemo>
sure, that's cool
<Venemo>
btw, can it also visualize changes between pcb revisions?
<Joerg-Neo900>
and a git ls or whatever the name
<Joerg-Neo900>
nope, alas not (yet)
<Joerg-Neo900>
viewing PCBs is a PITA even without trying to visualize diffs
<Joerg-Neo900>
... for multilayer PCB
<Venemo>
very true
<Venemo>
well, I could imagine a side-by-side layer-by-layer comparison, with the unmodified parts simply faded out
<Joerg-Neo900>
I could think of flickering diffs
<Joerg-Neo900>
but, as said, for multilayer you learn little by simplified display like e.g. one layer only
<Joerg-Neo900>
what you need is a massively anamorphotic display with layer thickness boosted to maybe factor 50, then highlight connected copper traces across all layers and vias
<Joerg-Neo900>
with 3D view
<Joerg-Neo900>
basically a 3D redering of the PCB with thickness changed from 0.8 to 80mm, with FR transparent
<Joerg-Neo900>
and semi-transparent traces and vias
<Joerg-Neo900>
in different colors for each layer
<Joerg-Neo900>
and highlight for any arbitrary particular network
<Joerg-Neo900>
then rotate and shift and uoom in/out with mouse and some more keys
<Venemo>
whoah
<Joerg-Neo900>
I *guess* most recent kicad is almost there
galiven_ has joined #neo900
<Venemo>
well, I didn't think of such complications for the diff view
<Joerg-Neo900>
the diggs would be flicker in that
<Joerg-Neo900>
diffs
<Venemo>
aha
<Venemo>
well, it sounds overkill
<Joerg-Neo900>
like fast tab clicks in eeshow
<Joerg-Neo900>
s/over// and you're genuinely at gerbviewers at large ;-)
<Joerg-Neo900>
I *think* eeshow already has the tools to visualize diffs between any two bmp/whatever pictures, be it schematics or one layer of a pcb
galiven has quit [Ping timeout: 240 seconds]
<Joerg-Neo900>
the default mode does graphical diff
<Joerg-Neo900>
you just need to skip the eeschema rendering and start with images right away as input data
<Joerg-Neo900>
after all aiui that is what the predecessor of eeshow did
<Joerg-Neo900>
so yes, that should be easily implemented or even already possible to show diffs of a single layer, with multople layers on consecutive pages which you browse by pgup/dwn
<Joerg-Neo900>
ask wpwrak ;-)
ghs39ghsggf has joined #neo900
ghs39ghsggf has left #neo900 [#neo900]
ubizzy has joined #neo900
<wpwrak>
image diff is in gfx/diff.c. good luck ! :)
<Joerg-Neo900>
can't you load giffs or bmp or whatever?
<Joerg-Neo900>
just bypassing the eeschema rendering
<wpwrak>
you could develop the code to use other sources. of course, if you just want an image diff, you could just use imagemagick
<Joerg-Neo900>
or that :-)
<Joerg-Neo900>
not as convenient as eeshow
<wpwrak>
compare -metric AE a.png b.png out.png
<Joerg-Neo900>
well, that's a static file aiui
<wpwrak>
oh, i used this to debug eeshow. so .. :)
infobot has joined #neo900
<Joerg-Neo900>
~sucker
<infobot>
hmm... sucker is something you shouldn't tell in public
<Joerg-Neo900>
~useless
* infobot
starts crying and hides from joerg-neo900 in the darkest corner of the room. :(
<Joerg-Neo900>
fun to insult the bot
<Venemo>
:D
<Joerg-Neo900>
tmepted to tell her ~die but from me she takes it literally
<Joerg-Neo900>
[2017-06-14 Wed 18:35:57] <sicelo> perfect .. bot gone when i need it
<Joerg-Neo900>
Venemo: indeed you could prepare your own gerber diff with a lengthy script employing imagemagick tools, and a picture viewer that allows fast diashows
<Venemo>
nah
<Venemo>
sounds like something worthwhile, but don't have time unfortunately
<Venemo>
anyway thanks for everything
<Venemo>
I gotta rush off, ttyl :)
<Joerg-Neo900>
yw, don't miss to spread the word! :-)
<Joerg-Neo900>
Venemo: >>Eeshow has evolved from the much older schhist project at Qi-Hardware that provided a collection of scripts that used Eeschema (with the help of Wolfgang Spraul's command-line patches) to render schematics to bitmap images and then to compare these, producing a large HTML table. << (from http://neo900.org/stuff/eeshow/) - does that give you ideas? you wouldn't need to make the scripts from scratch
cc___ has quit [Ping timeout: 240 seconds]
<Joerg-Neo900>
just s/eeschema/pcbnew|gerbview/
xes_ has joined #neo900
xes has quit [Ping timeout: 240 seconds]
xes__ has joined #neo900
xes_ has quit [Ping timeout: 246 seconds]
<Joerg-Neo900>
ooh wait, nit from this account :-)
<Joerg-Neo900>
~die
* infobot
takes two shots to the head and crumples to the ground, lifeless.
<Joerg-Neo900>
two!!
xes__ has quit [Read error: Connection reset by peer]
xes__ has joined #neo900
xes has joined #neo900
xes__ has quit [Ping timeout: 240 seconds]
xes_ has joined #neo900
xes has quit [Ping timeout: 260 seconds]
jonsger has joined #neo900
xes__ has joined #neo900
xes_ has quit [Ping timeout: 240 seconds]
xes has joined #neo900
xes__ has quit [Ping timeout: 246 seconds]
jonsger has quit [Ping timeout: 240 seconds]
xes_ has joined #neo900
xes has quit [Ping timeout: 255 seconds]
xes__ has joined #neo900
xes_ has quit [Ping timeout: 260 seconds]
xes has joined #neo900
xes__ has quit [Ping timeout: 255 seconds]
ossguy has quit [Quit: leaving]
Pali has quit [Ping timeout: 268 seconds]
err0r3o3 has joined #neo900
<Venemo>
Joerg-Neo900: sure I'll spread the word, I'll show eeshow around the lab, there are guys who will find it very useful