0
I Use This!
Inactive

Commits : Listings

Analyzed 1 day ago. based on code collected 1 day ago.
Jul 22, 2024 — Jul 22, 2025
Commit Message Contributor Files Modified Lines Added Lines Removed Code Location Date
* Modified: Now if the terminal have fixed encoding the scroll bars uses the original chars. That's because we can't trust in any redefined value. More... almost 24 years ago
* Added: A function to know if the terminal have a fixed encoding and we shouln't offer to the user recoding options. Boolean codePageVariable() More... almost 24 years ago
* Added: Two functions to compat layer library to get the high and low nibble of 16 bits values (according to the endianess of the platform). More... almost 24 years ago
* Fixed: Some times after a suspend/resume of the keyboard ttyname failed to return the name of the terminal (who knows why?). Now if the name was already known and the new calls fails the old information is used. It was part of an Andris patch that I applied but not enabled because I have doubts about a line that now I left commented. More... almost 24 years ago
* Modified: I cleaned the writeBlock() function of tscreen.cc. It became a mess. I removed the C_, C__ and C___ macros and the __SAVECURSOR_OK option. Also removed some redundant checks and reindented the code. * Fixed: Now writeBlock() is Endian safe. I got reports from David Nock <[email protected]> who was working in a Linux PPC. I used code ideas from the version that José Angel Sánchez Caso (JASC) <[email protected]> sent me but I couldn't find time to merge. More... almost 24 years ago
* Modified: I cleaned the writeBlock() function of tscreen.cc. It became a mess. I removed the C_, C__ and C___ macros and the __SAVECURSOR_OK option. Also removed some redundant checks and reindented the code. * Fixed: Now writeBlock() is Endian safe. I got reports from David Nock <[email protected]> who was working in a Linux PPC. I used code ideas from the version that José Angel Sánchez Caso (JASC) <[email protected]> sent me but I couldn't find time to merge. * Added: Two functions to compat layer library to get the high and low nibble of 16 bits values (according to the endianess of the platform). * Fixed: Some times after a suspend/resume of the keyboard ttyname failed to return the name of the terminal (who knows why?). Now if the name was already known and the new calls fails the old information is used. It was part of an Andris patch that I applied but not enabled because I have doubts about a line that now I left commented. More... almost 24 years ago
* Modified: [Linux] A lot of low level details in the TScreen class: * Added: An internal variable TerminalType to hold the type of terminal, it can be: VCSA_TERMINAL The Linux console with access to /dev/vcsa*. LINUX_TERMINAL The Linux console without access to /dev/vcsa* GENER_TERMINAL A generic terminal. * Modified: Now LINUX_TERMINAL sets the terminal to use G1 charset and sets the G1 charset to the screen font (user defined). It means that now TV applications that doesn't have access to /dev/vcsa* have access to all symbols anyways. I had to do some tricks to get some control characters printed. I used the UTF-8 mode that Linux have (unicode). So now the fonts loaded by the user will be used (I can see my accents correctly). * Fixed: The 127 and 128+27 codes are control characters and shouldn't be printed. It deformed the "ASCII" window. * Fixed: After suspend/resume the calls to endwin/doupdate messed up the selected charset. I think that's a bug in ncurses. It was noted when you suspended to a shell and resumed (frames damaged). Now the library ensures the right charset is selected after doupdate(). * Modified: Now RestoreScreen() (called by suspend) clears the screen if the user screen can't be restored. That's good for calling a shell and is less confusing than just do nothing. * Modified: Now when using GENER_TERMINAL only the frames are remapped from PC437 letting some ISO Latin 1 accents alive. That's not perfect but is better than previous behavior. I have ideas about how to enhance it. More... almost 24 years ago
* Added: Now the install target is separated in three parts: install-headers, install-static and install-dynamic. The install target means the three, but you can just install the dynamic library. It helped me a lot while testing things with the editor. More... almost 24 years ago
* Fixed: An error in the config.pl that put libintl.a for Linux target. More... almost 24 years ago
*** empty log message *** More... almost 24 years ago
* Added: A function to get the name of the shell to the compat layer library. More... almost 24 years ago
* Modified: Disable the sched_yield() method in yieldpro.c. I don't know if it's working, I guess so, but top says the applications uses 99% of the CPU. I guess that's because I only running this application (all the rest are sleepping) and hence Linux doesn't have anyone to yield the CPU and returns inmediatly. That's OK, but I'm sure users will complain because it gives the impression that we eat all the CPU. More... almost 24 years ago
* Added: A dummy libintl.a library. This is provided for people that have configured the library with international support but want to create one or more executables without it. More... almost 24 years ago
* Added: A dummy libintl.a library. This is provided for people that have configured the library with international support but want to create one or more executables without it. * Added: An explanation of it to the readme.txt file and also some explanation about the impact of libiconv.a. More... almost 24 years ago
* Added: Support for gettext 0.10.37 and newer that needs libiconv. I tested it with 0.10.39 and libiconv 1.7. The detection was enhanced and the examples are configured to use -liconv if needed. Thanks to Laurynas for ideas and some Perl patches. More... almost 24 years ago
* Modified: Now the yieldproc.c module uses the POSIX sched_yield function if available. It must be tested to see if that's good. Sugested by Robert Varga <[email protected]>. More... almost 24 years ago
* Modified: The debian files. Mainly to reflect the new dependencies and warn users about doubts in Borland's license. From Ivan Baldo. * Fixed: [Linux] The dynamic library didn't recompile the compat library with -fPIC. Report and ideas by Ivan Baldo. More... almost 24 years ago
* Fixed: We wasn't including the compat library as relocatable stuff. More... almost 24 years ago
* Cambios de Ivan referidos a nuevas versiones de las dependencias y warnings sobre las reservas en la distribución del paquete. More... almost 24 years ago
* Fixed: The bug mentioned in the last check in. Now when looking for the parent's console we stop if any parent isn't attached to a terminal. It works for my X because none of the X terminals I have are attached to a terminal, but we must see if that's enough. More... almost 24 years ago
* Fixed: Warning about unused variable in gkeyli.cc when HAVE_KEYSYMS was not enabled. Patch from Robert. More... almost 24 years ago
* Fixed: ticks.h used NULL without including stdio.h. Patch from Robert. More... almost 24 years ago
* Fixed: The terminal attributes saved in startcurses were not restored in stopcurses (but in the destructor) making assymetric the calls. Patch by Robert. * Modified: Now sys/perm.h is used and asm/io.h doesn't need to be protected from sys/io.h. Robert hint. * Added: Now if the library isn't in a virtual console the library tries harder to find out if this is situation is because the parent spawned the program in a terminal like a pts/x. If this is the case the library finds what terminal owns the parent and uses it. This is very nice to bypass programs like Midnight Commander, but have a serious drawback that I must fix in the next minutes: X is also bypassed ;-). Patch from Robert. More... almost 24 years ago
* Fixed: The terminal attributes saved in startcurses were not restored in stopcurses (but in the destructor) making assymetric the calls. Patch by Robert. * Modified: Now sys/perm.h is used and asm/io.h doesn't need to be protected from sys/io.h. Robert hint. * Added: Now if the library isn't in a virtual console the library tries harder to find out if this is situation is because the parent spawned the program in a terminal like a pts/x. If this is the case the library finds what terminal owns the parent and uses it. This is very nice to bypass programs like Midnight Commander, but have a serious drawback that I must fix in the next minutes: X is also bypassed ;-). Patch from Robert. * Fixed: ticks.h used NULL without including stdio.h. Patch from Robert. * Fixed: Warning about unused variable in gkeyli.cc when HAVE_KEYSYMS was not enabled. Patch from Robert. More... almost 24 years ago
* Added: The .mak files to the Linux distribution. Released as beta. More... almost 24 years ago
* Added: Makefiles for all the examples. They are created/updated when the package is created and updated by the configure script to be adapted to the target system. Documented. * Fixed: Various compilation errors in the examples. More... almost 24 years ago
* Added: The Dialog Editor to the distribution and mentioned in the readme. * Added: A .po file containing all the messages that can be internationalized in Turbo Vision. More... almost 24 years ago
* Added: A Dialog Editor created by Warlei Alves, he can't follow this project so now it will be part of the examples directory. More... almost 24 years ago
* Added: A check in TCluster::draw to avoid buffer overflows. I don't know if it is really needed but a user asked for it and is better to double check this nasty details. Sent by Warlei Alves. More... almost 24 years ago
* Added: A check in TCluster::draw to avoid buffer overflows. I don't know if it is really needed but a user asked for it and is better to double check this nasty details. Sent by Warlei Alves. * Added: A Dialog Editor created by Warlei Alves, he can't follow this project so now it will be part of the examples directory. More... almost 24 years ago