One other oddity. openSUSE has a libcairo, libgobject and a
libcairo-gobject.
In cairo deps build.jl I can see references to cairo and gobject, but not
to the third. Just looks a bit odd.
On Sunday, 10 April 2016 12:25:12 UTC-4, Colin Beckingham wrote:
>
> Did not work.
> I added an entry for tcl and tk on lines 6,7 and ran Pkg.build on Tk.
> Then I ran Pkg.update() where it refused to update Tk due to dirty which
> is good, it confirms I edited the right file.
> For reference here is the full error on "using Tk"
>
> INFO: Recompiling stale cache file /home/colin/.julia/lib/v0.5/Cairo.ji
> for module Cairo.
> WARNING: Method definition set_source(Cairo.CairoContext,
> Cairo.CairoSurface) in module Cairo at
> /home/colin/.julia/v0.5/Cairo/src/Cairo.jl:526 overwritten at
> /home/colin/.julia/v0.5/Cairo/src/Cairo.jl:546.
> ERROR: LoadError: LoadError: UndefVarError: _jl_libgobject not defined
> in init() at /home/colin/.julia/v0.5/Tk/src/tkwidget.jl:36
> in include(::ASCIIString) at ./boot.jl:234
> in include_from_node1(::ASCIIString) at ./loading.jl:417
> in include(::ASCIIString) at ./boot.jl:234
> in include_from_node1(::ASCIIString) at ./loading.jl:417
> in eval(::Module, ::Any) at ./boot.jl:237
> [inlined code] from ./sysimg.jl:11
> in require(::Symbol) at ./loading.jl:348
> in eval(::Module, ::Any) at ./boot.jl:237
> while loading /home/colin/.julia/v0.5/Tk/src/tkwidget.jl, in expression
> starting on line 454
> while loading /home/colin/.julia/v0.5/Tk/src/Tk.jl, in expression starting
> on line 37
>
> I will keep looking for other stray paths.
>
> On Sunday, 10 April 2016 11:51:18 UTC-4, Tim Holy wrote:
>>
>> Yes, probably relevant. Try adding it and then say `Pkg.build("Tk")`.
>> Assuming
>> that succeeds, then `using Tk` should have a high probability of working.
>>
>> --Tim
>>
>> On Sunday, April 10, 2016 08:28:29 AM Colin Beckingham wrote:
>> > FWIW, my tcl library on openSUSE lives in "/usr/lib64/libtcl8.6.so"
>> which
>> > does not match any of the entries in line 6 of file
>> > Tk.jl/blob/master/deps/build.jl. Relevant?
>> > I have noted in a couple other instances that differences in paths seem
>> to
>> > be important, unless Julia decides to load and compile own versions of
>> > tools.
>> >
>> > On Sunday, 10 April 2016 08:38:00 UTC-4, Colin Beckingham wrote:
>> > > Ok, I copied and pasted these lines including the "using" commands
>> and the
>> > > "using Tk" threw the subject line error.
>> > > I may have muddied the water a bit by not noticing that the first
>> time the
>> > > "using Tk" is issued the error appears, but if the same command is
>> > > repeated
>> > > in the same REPL session it is silent and looks reassuring, while in
>> point
>> > > of fact it is not.
>> > > So the using Tk is not happy.
>> > >
>> > > I have scanned the openSUSE Tk packages and as far as I can see the
>> > > obvious ones (and other non-obvious packages as a trial balloon)
>> should be
>> > > in place.
>> > >
>> > > On Sunday, 10 April 2016 07:46:05 UTC-4, Tim Holy wrote:
>> > >> Seems quite mysterious. To debug, the next thing I'd do is
>> > >>
>> > >> julia> cd(Pkg.dir("ImageView", "src"))
>> > >>
>> > >> and then copy/paste the code that's in src/ImageView.jl into the
>> REPL.
>> > >> (Probably just the part that's inside the `module ImageView ...
>> end`.)
>> > >> Presumably you should get error(s), and the line that triggers the
>> error
>> > >> will
>> > >> be a valuable clue.
>> > >>
>> > >> Best,
>> > >> --Tim
>> > >>
>> > >> On Saturday, April 09, 2016 10:07:18 PM Colin Beckingham wrote:
>> > >> > Thanks. Tk is loaded, not Gtk.
>> > >> > Resolved the error in Winston with a Pkg.checkout and build.
>> > >> > I did a Pkg.status("Tk") and this responds without error, version
>> Tk
>> > >>
>> > >> 0.3.7
>> > >>
>> > >> > Also did status on Winston, ImageView, Images, Cairo and all
>> respond
>> > >> > without error.
>> > >> > But still the error, and view() is not available.
>> > >> >
>> > >> > On Saturday, 9 April 2016 20:34:46 UTC-4, Tim Holy wrote:
>> > >> > > Try "using Winston" and then whos(); see if Gtk is loaded (or if
>> it's
>> > >>
>> > >> Tk).
>> > >>
>> > >> > > --Tim
>> > >> > >
>> > >> > > On Saturday, April 09, 2016 11:34:43 AM Colin Beckingham wrote:
>> > >> > > > I tried Pkg.rm("PyPlot") which put PyPlot in the additional
>> > >>
>> > >> packages
>> > >>
>> > >> > > group,
>> > >> > >
>> > >> > > > but the error persists. If there is something else I should
>> do, by
>> > >>
>> > >> all
>> > >>
>> > >> > > > means give a hint.
>> > >> > > >
>> > >> > > > I thought I found an older note which referred to ImageView
>> and
>> > >>
>> > >> Gtk,
>> > >>
>> > >> > > which
>> > >> > >
>> > >> > > > is why I mentioned it.
>> > >> > > >
>> > >> > > > Running through the require list for ImageView I have no
>> problems
>> > >>
>> > >> using
>> > >>
>> > >> > > any
>> > >> > >
>> > >> > > > of them except Winston, which produces
>> > >> > > > "ERROR: LoadError: syntax: space before "(" not allowed in
>> > >> > > > "AbstractVecOrMat{T} (""
>> > >> > > > I mention it in case it might be tangentially relevant.
>> > >> > > >
>> > >> > > > On Saturday, 9 April 2016 13:51:17 UTC-4, Tim Holy wrote:
>> > >> > > > > Not sure what's happening. Is it possible there's a conflict
>> with
>> > >> > >
>> > >> > > PyPlot?
>> > >> > >
>> > >> > > > > When I grep the source of ImageView, I don't find any
>> mention of
>> > >>
>> > >> gtk.
>> > >>
>> > >> > > > > --Tim
>> > >> > > > >
>> > >> > > > > On Saturday, April 09, 2016 10:06:52 AM Colin Beckingham
>> wrote:
>> > >> > > > > > Pkg.status()
>> > >> > > > > >
>> > >> > > > > > 31 required packages:
>> > >> > > > > > - BinDeps 0.3.21+ master
>> > >> > > > > > - Cairo 0.2.31
>> > >> > > > > > - Clustering 0.5.0
>> > >> > > > > > - ColorTypes 0.2.2
>> > >> > > > > > - DSP 0.0.11
>> > >> > > > > > - DataArrays 0.2.20
>> > >> > > > > > - DataFrames 0.7.0+ master
>> > >> > > > > > - Debug 0.1.6
>> > >> > > > > > - DecisionTree 0.3.12
>> > >> > > > > > - DimensionalityReduction 0.1.1
>> > >> > > > > > - Distances 0.3.0
>> > >> > > > > > - Docile 0.5.23
>> > >> > > > > > - FunctionalDataUtils 0.1.0
>> > >> > > > > > - GLM 0.5.0
>> > >> > > > > > - GZip 0.2.18
>> > >> > > > > > - HttpParser 0.1.1
>> > >> > > > > > - ImageMagick 0.1.3
>> > >> > > > > > - ImageView 0.1.19+ master
>> > >> > > > > > - Images 0.5.3
>> > >> > > > > > - IniFile 0.2.5
>> > >> > > > > > - JLD 0.5.9
>> > >> > > > > > - JSON 0.5.0
>> > >> > > > > > - LightXML 0.2.1
>> > >> > > > > > - MLBase 0.5.2
>> > >> > > > > > - PkgDev 0.1.0
>> > >> > > > > > - PyPlot 2.1.1+ master
>> > >> > > > > > - RDatasets 0.1.2+ master
>> > >> > > > > > - StatsBase 0.8.0
>> > >> > > > > > - Tk 0.3.7
>> > >> > > > > > - WAV 0.6.3
>> > >> > > > > > - Winston 0.11.13
>> > >> > > > > >
>> > >> > > > > > 29 additional packages:
>> > >> > > > > > - ArrayViews 0.6.4
>> > >> > > > > > - Blosc 0.1.4
>> > >> > > > > > - ColorVectorSpace 0.1.2
>> > >> > > > > > - Colors 0.6.3
>> > >> > > > > > - Compat 0.7.13
>> > >> > > > > > - Conda 0.1.9
>> > >> > > > > > - Dates 0.4.4
>> > >> > > > > > - Distributions 0.8.10
>> > >> > > > > > - FactCheck 0.4.2
>> > >> > > > > > - FileIO 0.0.4
>> > >> > > > > > - FixedPointNumbers 0.1.2 master
>> > >> > > > > > - FunctionalData 0.1.0
>> > >> > > > > > - Graphics 0.1.3
>> > >> > > > > > - HDF5 0.5.8
>> > >> > > > > > - HttpCommon 0.2.4
>> > >> > > > > > - Iterators 0.1.9
>> > >> > > > > > - LaTeXStrings 0.1.6
>> > >> > > > > > - MultivariateStats 0.2.2
>> > >> > > > > > - PDMats 0.4.1
>> > >> > > > > > - Polynomials 0.0.5
>> > >> > > > > > - PyCall 1.4.0
>> > >> > > > > > - Reexport 0.0.3
>> > >> > > > > > - SHA 0.1.2
>> > >> > > > > > - SIUnits 0.0.6
>> > >> > > > > > - SortingAlgorithms 0.0.6
>> > >> > > > > > - StatsFuns 0.2.0
>> > >> > > > > > - TexExtensions 0.0.3
>> > >> > > > > > - URIParser 0.1.3+ master
>> > >> > > > > > - Zlib 0.1.12
>> > >> > > > > >
>> > >> > > > > > On Saturday, 9 April 2016 12:48:49 UTC-4, Tim Holy wrote:
>> > >> > > > > > > I can't replicate this, whether with the tagged version
>> of
>> > >> > >
>> > >> > > ImageView
>> > >> > >
>> > >> > > > > > > (Pkg.free("ImageView")) nor with master. What does
>> > >>
>> > >> Pkg.status()
>> > >>
>> > >> > > say?
>> > >> > >
>> > >> > > > > > > --Tim
>> > >> > > > > > >
>> > >> > > > > > > On Saturday, April 09, 2016 07:49:44 AM Colin Beckingham
>> > >>
>> > >> wrote:
>> > >> > > > > > > > Using openSUSE Leap 42.1
>> > >> > > > > > > > I can add the package ImageView and even checkout the
>> > >>
>> > >> latest
>> > >>
>> > >> > > master
>> > >> > >
>> > >> > > > > and
>> > >> > > > >
>> > >> > > > > > > > build. Then on "using ImageView" the error
>> > >> > > > > > > > UndefVarError: _jl_libgobject not defined
>> > >> > > > > > > > pops out. I guess this has something to do with my
>> graphic
>> > >> > >
>> > >> > > backend,
>> > >> > >
>> > >> > > > > but
>> > >> > > > >
>> > >> > > > > > > not
>> > >> > > > > > >
>> > >> > > > > > > > sure what the next step is. I have seen other
>> references to
>> > >> > >
>> > >> > > problems
>> > >> > >
>> > >> > > > > > > with
>> > >> > > > > > >
>> > >> > > > > > > > libgtobject but nothing that seems to coincide with my
>> > >> > >
>> > >> > > situation.
>> > >> > >
>> > >> > > > > Same
>> > >> > > > >
>> > >> > > > > > > > problem on Julia v 4 and 5. I know from loading the
>> image
>> > >> > >
>> > >> > > package in
>> > >> > >
>> > >> > > > > > > Octave
>> > >> > > > > > >
>> > >> > > > > > > > that a backend GTK graphics engine is available. A
>> call to
>> > >> > >
>> > >> > > Pkg.add()
>> > >> > >
>> > >> > > > > > > > specifying GTK as an option does not seem to align
>> with any
>> > >> > >
>> > >> > > version
>> > >> > >
>> > >> > > > > of
>> > >> > > > >
>> > >> > > > > > > > .add() in an ImageView context.
>>
>>
