Re: [savi-users] running with geomview

[Jason M. <jas...@sr...>]

Hi Lloyd,

I made an assumption. In the README file, under "4. INSTALLATION INSTRUCTIONS FOR SAVI" references are made to "$SAVI". The line "we shall refer to the directory originally containing this README file, the root of the SaVi tree, as $SAVI." does NOT instruct the user to create a env var. But my brain made that leap anyway. 

Best,
Jason




On 32016, 2:00 AM, "llo...@ya..." <llo...@ya...> wrote:

>Jason,
>
>I'm curious. What prompted you to create that variable?
>If the readme could be taken that way, it needs editing.
>
>(The whole point of the savi script is to figure out
>setting SAVI to point to the savi directory, and TCL_LIBRARY
>and TK_LIBRARY so that TCLTK will work, before launching
>the binary. There's some other stuff like setting flags
>for odd environments - Mac OS X 10.5's Tcl was not good -
>but that's pretty much it.
>
>Setting those crossplatform makes the script
>a little complex; if you set all three yourself, you
>could probably run the binary directly, but since the
>binary needs support files, not sure why you'd want to.)
>
>SaVi 1.4.9a is out with the build error addressed - other
>mail.
>
>regards
>
> 
>Lloyd Wood
>http:savi.sf.net
>
>----- Original Message -----
>From: Jason Mellein <jas...@sr...>
>To: "llo...@ya..." <llo...@ya...>; "sav...@li..." <sav...@li...>
>Sent: Sunday, 20 March 2016, 7:16
>Subject: Re: [savi-users] running with geomview
>
>I built the unreleased code. It's working great now. But it turns out the key problem was I had created an shell environment variable SAVI with the canonical path to the savi directory. So I think it was looking for UsersmedevsaviUsersmedevsavi+ADs- which, of course, doesn't exist. 
>
>I deleted that env var and now geomview -run .savi works like a charm.
>
>Thanks for all the help+ACE-
>
>
>
>On 31916, 1:42 AM, +ACI-lloyd.wood+AEA-yahoo.co.uk+ACI- +ADw-lloyd.wood+AEA-yahoo.co.uk+AD4- wrote:
>
>+AD4-Jason,
>+AD4-
>+AD4-(Not sure what's going on with your message characters - smart quotes? Take a look below at your message...)
>+AD4-
>+AD4-the savi1.4.9.geomview file is described in savi1.4.9README, section 5, which is accessible from SaVi's Help menu. The only editing I can think there of is absolute path, which would be consistent with the problems with Geomview path handling (needing .savi instead of savi, not finding file).
>+AD4-
>+AD4-As you will have seen from my emailed output and from running SaVi standalone, SaVi always prints out a number of messages to stderr and the console when it launches. Do those messages appear when you launch SaVi from Geomview? Do those messages appear when you launch
>+AD4-cd savi1.4.9
>+AD4-geomview -run .savi
>+AD4-? Do you get an added file-not-found message when you try e.g.
>+AD4-
>+AD4-geomview -run .savi .dataglobalstar.tcl
>+AD4-? If no messages, SaVi is not even getting launched. What's printed to console is important.
>+AD4-
>+AD4-
>+AD4-(Note that you don't load Geomview .oogl files into SaVi yourself. .tcl script files creating the satellites or setting colours, certainly. Geomview can display and parse static .oogl files.)
>+AD4-
>+AD4-
>+AD4-Another point to consider is that your compiler is clang, not gcc. (I've tested building SaVi with clang up to 3.1 on borrowed Macs with recent Mac OS X, but not geomview. Because of the amount of junk I'd have to install.) And Fink's geomview will have been compiled with gcc, quite a while back. That might explain filepath handling, too+ADs- Mac OS X has gone weirder and further from its BSD roots, the Fink package was built several years ago. Still happily using both on Mac OS X 10.6.8.
>+AD4-
>+AD4-
>+AD4-https:support.enthought.comhcen-usarticles204469410-OS-X-GCC-Clang-and-Cython-in-10-9-Mavericks
>+AD4-
>+AD4-I don't know whether that causes problems+ADs- for all I know it's the Fink packaging, and I would suggest either:
>+AD4-a. building Geomview 1.9.5 from source with
>+AD4-cd geomview1.9.5
>+AD4-.configure
>+AD4-make
>+AD4-to eliminate Fink's package as a concern, though dependencies (especially the Latex stuff because you have to build documentation to build Geomview - my Cygwin page has a list of needed things for people who build on that) can be annoying. I've built geomview with opengl support under Fink previously, but that was a while back.
>+AD4-
>+AD4-
>+AD4-b. building both SaVi and Geomview with Fink's version of gcc, which is likely already installed, instead of clang?
>+AD4-
>+AD4-
>+AD4-b. Running Debian under an installation of VirtualBox and just click-installing both from Debian's library. Slightly older version of SaVi, but that gets the 3D running, and you can then build SaVi 1.4.9 as before using the packaged geomview there, with gcc. Gives you confidence that it all works as described before going back to Mac OS X.
>+AD4-
>+AD4-
>+AD4-The compilation error you describe is the one I fixed earlier this week to remove a warning I noticed with gcc (resulted from a code change to prevent a different warning that some linkers hard failed on), but that doesn't matter and doesn't affect startup or any functionality once compiled. Aborting compilation here on imo a minor warning is an artefact of clang being newly fussy, I think, since I've yet to see gcc do the same. C and compilers are traditionally very forgiving of this kind of thing+ADs- gcc warns because of all the flags I set, but lets this one slide. But regardless, breaking compilation in a release, even on a rare compiler, should need a release fix. That bug is fixed, with some new features, in the unreleased code (19 March tarball) at
>+AD4-http:personal.ee.surrey.ac.ukPersonalL.WoodsoftwareSaVisrcunreleased
>+AD4-
>+AD4-
>+AD4-The only other things to note:
>+AD4-
>+AD4-64-bit Cygwin does have trouble with SaVi and Geomview communicating via the pipe - noted in BUGS. It seems extremely unlikely that Mac OS X is now affected the same way, however, given that both have worked well for years, Linux works, 32-bit Cygwin works...
>+AD4-
>+AD4-
>+AD4-installing ActiveTcl is way better and more reliable in interface than the obsolete Tcl provided with Mac OS X (install dmg package, uncomment the line in srcMakefile+AF8-defs+AF8-macosx, recompile SaVi, check about SaVi and TclTk version - noted in savi1.4.9BUGS) but that should have no bearing on the savigeomview piping interaction problem you face. Whether SaVi is actually being launched at all and able to print statusorbital model used to console is much more important.
>+AD4-
>+AD4-
>+AD4-hope this helps,
>+AD4-
>+AD4- 
>+AD4-Lloyd Wood
>+AD4-http:savi.sf.net
>+AD4-
>+AD4-
>+AD4-
>+AD4------ Original Message -----
>+AD4-From: Jason Mellein +ADw-jason.mellein+AEA-sri.com+AD4-
>+AD4-To: +ACI-lloyd.wood+AEA-yahoo.co.uk+ACI- +ADw-lloyd.wood+AEA-yahoo.co.uk+AD4AOw- +ACI-savi-users+AEA-lists.sourceforge.net+ACI- +ADw-savi-users+AEA-lists.sourceforge.net+AD4-
>+AD4-Sent: Saturday, 19 March 2016, 15:38
>+AD4-Subject: Re: +AFs-savi-users+AF0- running with geomview
>+AD4-
>+AD4-
>+AD4-
>+AD4AKw-AD4APg- panel, you should be able to click on it to launch SaVi. If SaVi doesn't launch from there, you need to edit the location of SaVi in the dot file and rerun Geomview.
>
>+AD4-
>+AD4-This may be something to focus on. I click on it but it does a weird switch like its trying to do something but returns to the plain +-ACI-SaVi+-ACI- entry in the external modules list. It seemed to me that Geomview got what it needed because it got the CAPS right. But nothing opens. The SaVi entry quickly switches to something else and back again so I can't tell what it's trying to do. What dot file are we talking about here? And where is it located. I can try an edit.