I'm running 10.8.4 and Xcode 4.6.3. I didn't modify the script, and I'm sure I have all the dependencies because I did successfully install TILP before using Terminal.

I tried the new windows version, now it's saying it is missing libticables2-7.dll :<

I installed TILP 1.18 from Lionel's website, and I am unable to change calculators or cables. The interface appears like this, and neither left- nor right-clicking in the blank part of the left half does anything. The platform is Windows 7 x64, and I chose to install GTK with TILP, so it grabbed whatever GTK version TILP requests.

Looks like the menus labels got messed up ?
The 2nd or 3rd one should bring up setup options.

But I'll let Lionel reply....

We already discussed about it on IRC, but nothing conclusive. I don't remember receiving such a report before, so chances are that there's something with Kerm's computer, but what ?
Jonimus suggested a corrupt GTK+ theme.

Something similar happens with right-clicking: the menus are all horizontal lines with a few scattered checkmarks and no text. Thanks for the response.

Related: I can't seem to get any setting to show my TI-Nspire CX (using the DirectLink cable, presumably); am I doing something wrong? Have I selected the wrong cable? Do I also need to install TI's student software?

Edit: Many thanks to tifreak8x for his tutorial here: www.cemetech.net/forum/viewtopic.php?t=10535

In both panes, the menus triggered by right-clicking contain several icons, but even if the icons were missing, I'd imagine that they should have text nevertheless.
Does the .tilp.log file in your Windows home directory say something relevant ?

tifreak's tutorial is clearly required for Windows 8 x64, but definitely not for Windows 7, unless Microsoft changed something recently. So far, I've never had to execute Zadig on any of my real / virtual Windows computers.
For the purposes of installing TILP II, installing TI-Connect and/or TINC(L)S only makes matters harder: users then have to install the filter driver, as described in the Windows README
The other, saner OS don't have such fundamental limitation as requiring a driver for vendor-specific USB devices in the first place, and then having to put an additional layer of complexity for coping with multiple drivers for a given piece of hardware (since libticables and TI's software can't use the same drivers, for licensing and functionality reasons)...

I was going to suggest using the filter driver installation procedure no matter what, it could have helped for some reason.

FWIW, I've felt motivated enough to download the Windows 10 Technical Preview, installed it in a VM (it's not immediately obvious how to do it without a Microsoft account), and installed http://tiplanet.org/beta/setup.exe into it.
Unsurprisingly, out of the box, my 89T wasn't detected, since the driver doesn't have a proper CAT file for Win 8 and above. However, after using Zadig to install the latest libusb-win32 driver for the 89T, communication with the 89T worked just fine
That is, dirlist and screenshot worked... but the fact that those work shows that there's no driver-related issue, at least.
That's good news, even if there's no guarantee that anything that works on the TP will work on the final release.

Unsurprisingly as well, TILP's UI worked properly. Kerm's computer remains a mystery flare...

I should have made a snapshot of the VM before installing TILP and using Zadig. I'll now have a harder time checking the procedure for installing TILP after TI-Connect / TINCLS...
Without reinstalling Windows entirely, can I be confident that some less heavy-handed approaches, starting with unselecting the driver manually, will get me close enough to the OOTB situation ? I'd guess so, since unselecting the driver is part of the fixing procedure we suggest users after they plugged their calculators / cables before installing TI-Connect / TINC(L)S / TILP.

I'll have to update the Win32 README to mention Zadig, too.

New version with bugfixes and improvements, as usual. Let's mention:
* fewer memory errors => fewer crashes;
* addition of extern "C" guards in public headers, so that libti* can be used from C++ programs... not that anybody cares, but this is still the right thing to do
* OS transfer fixes for 83+/84+, by Jonimus;
* support for TI-Z80 OS / FlashApp files in non-standard Intel Hex format (LF, CR), by myself for SirCmpwn;
* new ticables_supported_cables(), ticalcs_supported_calcs(), ticables_get_device() and ticables_set_device() APIs.

Known bugs:
* 84+ DirectLink ROM dumping doesn't work anymore. None of the recent commits should have triggered that...
* transferring Franc,ais or Espan~ol 84+ FlashApps triggers improper behaviour, due to encoding mishandling and old, insecure code.
* several harmful semicolons in multiple libticables files could theoretically trigger infinite loops. noticed during code inspection for other changes. Those bugs are many years old.

The usual links:
* Windows installer download: http://tiplanet.org/beta/setup.exe .
* Source code: https://github.com/debrouxl/tilibs + https://github.com/debrouxl/tilp_and_gfm .
* install script for anything non-Windows which is handled by TILP: https://github.com/debrouxl/tilp_and_gfm/blob/master/tilp/trunk/build/scripts/install_tilp.sh .

The TODO list contains items such as, but not limited to:
* finish introduction of a fifth libti* library, dubbed "libtiopers", for factoring code useful to multiple libti* clients. For now, there's pretty little in the new library, and disentangling is not that easy...
* finish GTK+3 compatibility patch;
* add a TILP menu entry for exiting the PTT mode (greyed out if not connected to a Nspire);
* make Nspire "ROM dumping" (OS dumping) work on all models, by duplicating BrandonW's Fron;
* minor installer and file encoding fixes.

I have started writing a single tutorial for installing TILP on Windows, Linux, MacOS X, gathering information from multiple places (TILP README, etc.)

EDIT in 2021: updated the link to the *nix install script.

On the tilibs repository ( https://github.com/debrouxl/tilibs ), there's some ongoing activity. Not the best thing I'm doing for my own sake (getting a rest and being productive)...
* the usual code review, consolidation, fixing and hardening work (multiple commits), quite a number of hours of work. Dozens of bugs (usually in error paths, but still) were pointed by static analysis tools, fixes in progress;
* improved 84+CSE support, by integrating a bunch of older patches by Benjamin Moody;
* initial support for '2015 models: 83PCE / 84+CE(-T) (no FlashApp support yet, experimental send OS code), 82A (as a crippled 84+; send OS code not tested, FlashApps not available). The probing code was improved;
* experimental support for enumerating > 1 level of folders on the Nspire. The corresponding UI (TILP) changes haven't been performed yet.
(some of those changes are available only on the "experimental" branch for now)

No updated public Windows build yet.

Since the previous post, I have added FlashApp send & recv support for the 83PCE & 84+CE, and updated the Windows beta build.

The send OS code is untested against a 84+CE, I need someone to perform the testing for me. TIA

The usual links:
* Windows installer download: http://tiplanet.org/beta/setup.exe .
* Source code: https://github.com/debrouxl/tilibs + https://github.com/debrouxl/tilp_and_gfm .
* install script for anything non-Windows which is handled by TILP: https://github.com/debrouxl/tilp_and_gfm/blob/master/tilp/trunk/build/scripts/install_tilp.sh .

EDIT in 2021: updated the link to the *nix install script.

I feel like I want to redo TILP in Cocoa because the GTK+ interface is somewhat confusing (for me, anyway).
Also, TI-84+ clock change has been broken for a while...

Go ahead if you wish

FYI, Glib is a hard dependency of the libti* family:
* implicit for libticonv, libtifiles and libticables, which return memory allocated through Glib and use glib's character conversion functions (the APIs abundantly leak this, especially libticonv's);
* explicit for libticalcs, whose API uses Glib types such as GNode, besides possible implicit dependencies.

Several APIs for abstracting out the memory allocator somewhat were already added. The libti* family is still missing at least:
* freeing counterparts to tifiles_error_get(), ticables_error_get(), ticalcs_error_get();
* freeing counterparts to a number of libticonv functions;
* freeing counterparts to tifiles_fext_dup(), tifiles_build_filename();
* reallocating counterparts to tifiles_ve_alloc_data(), tifiles_fp_alloc_data().
Such functions would help all clients of libti*.

Given that platform-specific UIs were used a long time ago for TILP, but proved to be a maintenance burden in the past, platform-specific UIs would certainly remain unofficial (unlike a cross-platform Qt UI) - but being unofficial doesn't prevent something from being useful to users


News to me, written down into the todo/wish list.
Was that through direct USB, or through the legacy I/O port ?

Direct USB.

In my Win 7 x64 VM, TILP II 1.18 beta now displays the same messed up top menu bar, and right-click menu entries, as Kerm screenshotted above. It's not just Kerm's computer

uyjulian: I can also reproduce your problem, under Linux with HEAD libticalcs but not on Windows with the 20150426 beta build, where clock synchronization works well, AFAICT.
Your message states that the problem is older, so I'm at a loss.

I have been unable to dump the rom for my TI-89T using TiLP on Windows 10. Other Asm/BASIC programs transfer properly (most of the time), but the calculator tunrs off when I try to start to dump the ROM.

I hear that there's a problem for you, but it works for me, through both the legacy I/O port and the direct USB port...
Could you post the contents of ~/.tilp.log, where ~ is the home directory of your user, usually C:\Users\<username> nowadays ?

Older? No, it's still there on commit c406a654ea9333f029383584c2553770d1b16cc4
By the way, I'm running Mac OS X.

I meant that the problem has been happening for a longer period of time than several weeks, according to your report
The fact that it somehow works on Windows, but not on Linux (with HEAD of experimental branch, which is quite a bit ahead of master) or MacOS X (with HEAD of master branch) makes no sense.

Noteworthy changes since the previous post:
* the send OS code was tested against a 84+CE by Adriweb, thanks;
* two fixes for the libusb 1.0 context teardown, the main one from Ben "TC01" Rosser and the smaller, but related one by myself;
* Benjamin Moody fixed remote variable deletion of archived variables on 84+ family DUSB and 89T. The error message made it sound like the feature was not supported, which probably explains why we didn't get more complaints, but that wasn't true ;
* I integrated the GrayLink fix from the patched libticables provided by TI-Edit. Peter Engels says that enabling DTR, for power supply purposes, fixes the operation of a third-party GrayLink, and that the fix makes the behaviour of libticables match the behaviour of the old TI-GraphLink software. IOW, the old behaviour of libticables was incorrect.
* as usual, I performed scattered fixes and cleanups, pretty much invisible to the end user, unless you're e.g. an user of FreeBSD or OpenBSD which compiles pre-release versions of libti*;
* I added more functions in the libraries, for symmetry and abstraction of the libraries' implementation, and for moving code originating from tilp and titools where it belongs. Those are invisible to the end user as well.

biohax17 and whoever else made third-party GrayLink-compatible cables: I need your help for testing your cables against this version of libticables
Although the new libticables behaviour is here to stay, for the reasons stated above, it would be unfriendly to break operation of your cables (though I doubt it's going to happen) in an official TILP release without prior notice.

An updated glimpse into the libti*/gfm/tilp todo/wish/bug list file, which is over 35 KB large, partially due to verbatim excerpts of e-mails:
* current development cycle:
- moving several more functions from clients to libti*;
- TILP UI changes for > 1 level of folders on the Nspire, I suppose;
- new functions for parsing data in TI certificate format, used on all TI graphing calculator series;
- Jonimus' driver-less support for BlackLink / $4 cable on Windows, based on bit-banging;
- other fixes and improvements, most of which invisible to end users.

* later development cycles:
- more code motion from clients to libti*, including a new "libtiopers" library;
- TCP client and TCP server cables;
- reworked, much improved packet logging and dissection;
- further improvement to USB calculator probing;
- portable, generic Nspire OS dumping (functionality known from BrandonW's Fron);
- add a TILP menu entry for exiting the PTT mode (greyed out if not connected to a Nspire);

The usual links:
* Windows installer download: http://tiplanet.org/beta/setup.exe .
* Source code: https://github.com/debrouxl/tilibs (experimental branch) + https://github.com/debrouxl/tilp_and_gfm (experimental branch).
* install script for anything non-Windows which is handled by TILP: https://github.com/debrouxl/tilp_and_gfm/blob/master/tilp/trunk/build/scripts/install_tilp.sh .

EDIT in 2021: updated the link to the *nix install script.