From: Tijl C. <ti...@ul...> - 2003-01-05 03:48:20
|
Hello to all TiLP developers, Today I finished porting TiLP to FreeBSD. A set of patches for each package is available at http://tijl.studentenweb.org/tilp/ Currently I only have the Grey TIGL working. I might take a look at the others when I finish my exams at the end of January. What follows are a few notes about the patch sets. ** General FreeBSD currently doesn't have <stdint.h>. I believe the upcoming 5.0 will support it though. To overcome this I encapsulated each #include <stdint.h> with: +#ifdef HAVE_STDINT_H #include <stdint.h> +#else +#include <inttypes.h> +#endif Most of the patches just do this. I realise this isn't really a nice structure to have in your code, so you may not want to apply these. I also added LIBS="$LIBS $INTLLIBS" in every ./configure. With this line each package can be compiled with gettext in libintl in case glib12 isn't available. For libticalcs and tilp this doesn't matter much since they depend on glib12 anyway. Next I noticed the ti*-config scripts had a bug when the --cflags option was given. They returned the dir without the -I prefix. ** libticables 3.5.3 I added a few lines to ./configure to define __FBSD__. These should be added to configure.ac though. I didn't because then I prevented using autoconf/automake/... during compilation. I wonder why you don't use any gcc predefined constants though. I believe either unix or __unix__ is defined on every unix platform. This is what I get on a FreeBSD box. tijl tijl% gcc -dM -E - < /dev/null #define __FreeBSD__ 4 #define __FreeBSD_cc_version 460001 #define __i386__ 1 #define __i386 1 #define __GNUC_MINOR__ 95 #define i386 1 #define __unix 1 #define __unix__ 1 #define __GNUC__ 2 #define __ELF__ 1 #define unix 1 tijl tijl% Further on, the serial devices on FreeBSD are named /dev/cuaa(0|1|2|3). The parallel ports are /dev/lpt(0|1|2), but I haven't added these yet, since the parallel cable isn't supported yet anyway. On another note, FreeBSD doesn't define O_SYNC. With O_FSYNC however, things work perfect. ** libtifiles 0.4.0 No system specific patches were necessary. <stdint.h> is used quite a lot though. ** libticalcs 4.2.8 In src/calc_def.h I replaced <glib-1.2/glib.h> with <glib.h>, since I believe the glib-1.2 part is system dependent and ./configure adds a proper -I(include) to CFLAGS anyway. The downside is that TiLP's configure script chokes when checking for <tilp/calc_def.h>, but this can be overcome by adding the proper -I(include) to CPPFLAGS. At the moment this is done automatically by a Makefile I made for TiLP's entry in the FreeBSD ports tree, but I can imagine it isn't so convenient when you have to run ./configure manually. Probably, some construction can be figured out, like adding a CPPFLAGS= line at the same place as LIBS= CFLAGS= lines currently in there somewhere. Related to this, I could comment out a redefinition of HAVE_TILP_TICALCS_H in TiLP's src/tilibs.h. ** tilp 6.00 pre15 On some lines in the configure script $(prefix) is used. This gave some problems so I replaced them by $prefix. Should there be any difference between these two? src/term_cb.c wants to include error.h, but there doesn't seem to be any. ** That should be it. I hope this can be of help to somebody. I'm not submitting anything to the FreeBSD ports tree yet as I first want to check a few more things, maybe take a look at the usb lib/driver and that's all for after exams. At the moment I'm just happy to be able to connect to my calc! Tijl Coosemans |
From: Julien B. <jb...@jb...> - 2003-01-05 11:19:20
|
Tijl Coosemans <ti...@ul...> wrote: > Hello to all TiLP developers, Hi Tilj, > Today I finished porting TiLP to FreeBSD. A set of patches for each > package is available at http://tijl.studentenweb.org/tilp/ Good news :) > Currently I only have the Grey TIGL working. I might take a look at > the others when I finish my exams at the end of January. It shouldn't be too problematic to enable support for the other cables. You'll need to figure out the low-level access routines, and that's it :) For the USB cable, you'll need the libusb which is ported to FreeBSD already. For the parallel cable, maybe we could use the libieee1284, although I'm not sure it would cover our needs. > ** General > > FreeBSD currently doesn't have <stdint.h>. I believe the upcoming 5.0 > will support it though. To overcome this I encapsulated each #include > <stdint.h> with: Argh too bad :/ stdint.h is part of ISO C99... > I also added LIBS="$LIBS $INTLLIBS" in every ./configure. With this This should already be done, if it isn't then it's a bug. > Next I noticed the ti*-config scripts had a bug when the --cflags > option was given. They returned the dir without the -I prefix. Needs to be fixed. > ** libticables 3.5.3 > > I added a few lines to ./configure to define __FBSD__. These should be > added to configure.ac though. I didn't because then I prevented > using autoconf/automake/... during compilation. I wonder why you don't > use any gcc predefined constants though. I believe either unix or > __unix__ is defined on every unix platform. This is what I get on a > FreeBSD box. We don't want to rely on GCC specific features, so that TiLP can be built easily on other platforms using other compilers (eg Tru64 with Compaq's CC -- which is crap, for what matters :)). > ** libtifiles 0.4.0 > > No system specific patches were necessary. <stdint.h> is used quite a > lot though. We should have a specific header file defining our own types using whatever is available on the system, preferably stdint.h, falling back to inttypes.h or etc... > ** libticalcs 4.2.8 > > In src/calc_def.h I replaced <glib-1.2/glib.h> with <glib.h>, since I *dung* /me applies a big LART to Romain > believe the glib-1.2 part is system dependent and ./configure adds a > proper -I(include) to CFLAGS anyway. The downside is that TiLP's > configure script chokes when checking for <tilp/calc_def.h>, but this > can be overcome by adding the proper -I(include) to CPPFLAGS. configure needs to be fixed wrt this. > At the moment this is done automatically by a Makefile I made for > TiLP's entry in the FreeBSD ports tree, but I can imagine it isn't so > convenient when you have to run ./configure manually. Probably, some > construction can be figured out, like adding a CPPFLAGS= line at the > same place as LIBS= CFLAGS= lines currently in there somewhere. The autoconf check for libticalcs needs to be fixed to include the output of ticalc-config, or something like this. > Related to this, I could comment out a redefinition of > HAVE_TILP_TICALCS_H in TiLP's src/tilibs.h. This include file should GO AWAY AND BURN IN HELL, please Romain... It's one of the reason why TiLP takes so long to build... > ** tilp 6.00 pre15 > > On some lines in the configure script $(prefix) is used. This gave > some problems so I replaced them by $prefix. Should there be any > difference between these two? ${prefix} would be better, needs checking & fixing. > src/term_cb.c wants to include error.h, but there doesn't seem to be > any. IIRC it was missing in one of the tarballs. > That should be it. I hope this can be of help to somebody. I'm not > submitting anything to the FreeBSD ports tree yet as I first want to > check a few more things, maybe take a look at the usb lib/driver and > that's all for after exams. At the moment I'm just happy to be able to > connect to my calc! We definitely need to find some motivation and go on a cleanup spree in TiLP. And I need to go back to the TiLP for Mac OS X port... Not speaking of TiEmu :) Oh, well... My TODO list for TiLP would be : - use packageconfig instead of ticalc-config & friends (easy) - cleanup the #include mess (not too difficult) - double check the autoconf macros for libti* (not too difficult) After that, we could go on with a GTK 2 port. Thoughts ? JB. -- Julien BLACHE <http://www.jblache.org> <jb...@jb...> |
From: Romain L. <ro...@ti...> - 2003-01-05 15:07:31
|
Hi Tilj, On Sun, Jan 05, 2003 at 12:19:12PM +0100, Julien BLACHE wrote: > Tijl Coosemans <ti...@ul...> wrote: > > > Hello to all TiLP developers, > > Hi Tilj, > > > Today I finished porting TiLP to FreeBSD. A set of patches for each > > package is available at http://tijl.studentenweb.org/tilp/ > > Good news :) > Yeah ! Great ! An attempt has been done one day but without any success. I am very happy you managed to do that... > > Currently I only have the Grey TIGL working. I might take a look at > > the others when I finish my exams at the end of January. > > It shouldn't be too problematic to enable support for the other > cables. You'll need to figure out the low-level access routines, and > that's it :) > > For the USB cable, you'll need the libusb which is ported to FreeBSD > already. > > For the parallel cable, maybe we could use the libieee1284, although > I'm not sure it would cover our needs. > > > ** General > > > > FreeBSD currently doesn't have <stdint.h>. I believe the upcoming 5.0 > > will support it though. To overcome this I encapsulated each #include > > <stdint.h> with: > > Argh too bad :/ stdint.h is part of ISO C99... > > > I also added LIBS="$LIBS $INTLLIBS" in every ./configure. With this > > This should already be done, if it isn't then it's a bug. Yes, some work for me :-( > > > Next I noticed the ti*-config scripts had a bug when the --cflags > > option was given. They returned the dir without the -I prefix. > > Needs to be fixed. > I recently migrated to pkg-config. The ti*-config scripts are still generated but pkg-config is more powerful. > > ** libticables 3.5.3 > > > > I added a few lines to ./configure to define __FBSD__. These should be > > added to configure.ac though. I didn't because then I prevented > > using autoconf/automake/... during compilation. I wonder why you don't > > use any gcc predefined constants though. I believe either unix or > > __unix__ is defined on every unix platform. This is what I get on a > > FreeBSD box. > > We don't want to rely on GCC specific features, so that TiLP can be > built easily on other platforms using other compilers (eg Tru64 with > Compaq's CC -- which is crap, for what matters :)). > Such as MSVC even if I stopped the Win32 port. > > ** libtifiles 0.4.0 > > > > No system specific patches were necessary. <stdint.h> is used quite a > > lot though. > > We should have a specific header file defining our own types using > whatever is available on the system, preferably stdint.h, falling back > to inttypes.h or etc... > > > ** libticalcs 4.2.8 > > > > In src/calc_def.h I replaced <glib-1.2/glib.h> with <glib.h>, since I > > *dung* > > /me applies a big LART to Romain What's LART ? > > > believe the glib-1.2 part is system dependent and ./configure adds a > > proper -I(include) to CFLAGS anyway. The downside is that TiLP's > > configure script chokes when checking for <tilp/calc_def.h>, but this > > can be overcome by adding the proper -I(include) to CPPFLAGS. > > configure needs to be fixed wrt this. > > > At the moment this is done automatically by a Makefile I made for > > TiLP's entry in the FreeBSD ports tree, but I can imagine it isn't so > > convenient when you have to run ./configure manually. Probably, some > > construction can be figured out, like adding a CPPFLAGS= line at the > > same place as LIBS= CFLAGS= lines currently in there somewhere. > > The autoconf check for libticalcs needs to be fixed to include the > output of ticalc-config, or something like this. > > > Related to this, I could comment out a redefinition of > > HAVE_TILP_TICALCS_H in TiLP's src/tilibs.h. > > This include file should GO AWAY AND BURN IN HELL, please Romain... Yeah but I do this due to this fucking problem with tilp/calc_def.h and glib (as you mentionned it above). > > It's one of the reason why TiLP takes so long to build... > > > ** tilp 6.00 pre15 > > > > On some lines in the configure script $(prefix) is used. This gave > > some problems so I replaced them by $prefix. Should there be any > > difference between these two? > > ${prefix} would be better, needs checking & fixing. > > > src/term_cb.c wants to include error.h, but there doesn't seem to be > > any. > > IIRC it was missing in one of the tarballs. > > > That should be it. I hope this can be of help to somebody. I'm not > > submitting anything to the FreeBSD ports tree yet as I first want to > > check a few more things, maybe take a look at the usb lib/driver and > > that's all for after exams. At the moment I'm just happy to be able to > > connect to my calc! > > We definitely need to find some motivation and go on a cleanup spree > in TiLP. And I need to go back to the TiLP for Mac OS X port... JB: as I told you, some guys on the ti-fr forum are using the OS-X version and are happy with it... > > Not speaking of TiEmu :) Oh, well... > Oops ... > > My TODO list for TiLP would be : > - use packageconfig instead of ticalc-config & friends (easy) JB: it has been done ! Maybe I forgotten to tell you ? > - cleanup the #include mess (not too difficult) > - double check the autoconf macros for libti* (not too difficult) > > After that, we could go on with a GTK 2 port. > > Thoughts ? > > JB. > Well, you are welcome to modify TiLP so that it can be ported on FreeBSD. In a first step, I will try and merge your modifications... It seems that my tree is more recent that yours. Anyways, I'm going to use CVS given that I have an ADSL connection now. > -- > Julien BLACHE <http://www.jblache.org> > <jb...@jb...> > > > ------------------------------------------------------- > This sf.net email is sponsored by:ThinkGeek > Welcome to geek heaven. > http://thinkgeek.com/sf > _______________________________________________ > TiLP-devel mailing list > TiL...@li... > https://lists.sourceforge.net/lists/listinfo/tilp-devel > roms. -- Romain Lievin, aka 'roms' <ro...@ti...> The TiLP project is on <http://www.ti-lpg.org> "Linux, y'a moins bien mais c'est plus cher !" |
From: Julien B. <jb...@jb...> - 2003-01-05 15:28:51
|
Romain Lievin <ro...@ti...> wrote: >> > I also added LIBS=3D"$LIBS $INTLLIBS" in every ./configure. With t= his >>=20 >> This should already be done, if it isn't then it's a bug. > > Yes, some work for me :-( The autoconf check for libintl should do that. > I recently migrated to pkg-config. The ti*-config scripts are still gen= erated > but pkg-config is more powerful. Good ! You forgot to notify me... >> We don't want to rely on GCC specific features, so that TiLP can be >> built easily on other platforms using other compilers (eg Tru64 with >> Compaq's CC -- which is crap, for what matters :)). > > Such as MSVC even if I stopped the Win32 port. Yeah, but I was specificaly speaking of UNIX platforms, as, we all know it, everything else simply sucks :-) >> > In src/calc_def.h I replaced <glib-1.2/glib.h> with <glib.h>, sinc= e I >>=20 >> *dung* >>=20 >> /me applies a big LART to Romain > > What's LART ? A Luser Attitude Readjustment Tool. Which is, to say, any object you can use to hurt somebody with prejudice. :-) >> > Related to this, I could comment out a redefinition of >> > HAVE_TILP_TICALCS_H in TiLP's src/tilibs.h. >>=20 >> This include file should GO AWAY AND BURN IN HELL, please Romain... > > Yeah but I do this due to this fucking problem with tilp/calc_def.h and= glib > (as you mentionned it above). Nah, you use it to avoid #include'ing every needed header file in each source file, which results in a big header bloat. Now that the windows port is gone, we can get rid of it. >> We definitely need to find some motivation and go on a cleanup spree >> in TiLP. And I need to go back to the TiLP for Mac OS X port... > > JB: as I told you, some guys on the ti-fr forum are using the OS-X vers= ion and > are happy with it... How can they be happy with the buggy piece of shit it is ? There are obvious bugs, the USB cable support is buggy, and a couple of features are still lacking. >> Not speaking of TiEmu :) Oh, well... > > Oops ... I'm still waiting the bugfix to load a flash upgrade :-) (convert_tib_to_rom() IIRC) >> My TODO list for TiLP would be : >> - use packageconfig instead of ticalc-config & friends (easy) > > JB: it has been done ! Maybe I forgotten to tell you ? Yep. > It seems that my tree is more recent that yours. Anyways, I'm going to = use CVS > given that I have an ADSL connection now. Do not touch the CVS without asking me first, there's a tilp module that in fact resolves to tilp-osx. I need to remove that before. Importing the code for the libs into the CVS will require updating the OSX support for them, otherwise we'll have some license issue. Regarding the read_ico() functions, it is completely unportable. Get rid of it ASAP. JB. --=20 Ca fait d=E9j=E0 plusieurs semaines que j"essaie de correspondre par l" interm=E9diaire d"un groupe de discussion. Ca ne fait que deux jour que = je suis parvenu =E0 exp=E9dier un message -+- GM in GNU : Patience et longueur de temps.. -+- |
From: Romain L. <ro...@ti...> - 2003-01-05 15:42:27
|
Hi, > Yeah, but I was specificaly speaking of UNIX platforms, as, we all > know it, everything else simply sucks :-) > > It seems that my tree is more recent that yours. Anyways, I'm going to use CVS > > given that I have an ADSL connection now. > > Do not touch the CVS without asking me first, there's a tilp module > that in fact resolves to tilp-osx. I need to remove that before. I attempted but your cellular phone is not reachable.... > > Importing the code for the libs into the CVS will require updating the > OSX support for them, otherwise we'll have some license issue. > Really ? > Regarding the read_ico() functions, it is completely unportable. Get > rid of it ASAP. Why are theynon portable ? Ok, I will convert them... > > JB. > > -- > Ca fait d?j? plusieurs semaines que j"essaie de correspondre par l" > interm?diaire d"un groupe de discussion. Ca ne fait que deux jour que je > suis parvenu ? exp?dier un message > -+- GM in GNU : Patience et longueur de temps.. -+- > > roms. > ------------------------------------------------------- > This sf.net email is sponsored by:ThinkGeek > Welcome to geek heaven. > http://thinkgeek.com/sf > _______________________________________________ > TiLP-devel mailing list > TiL...@li... > https://lists.sourceforge.net/lists/listinfo/tilp-devel -- Romain Lievin, aka 'roms' <ro...@ti...> The TiLP project is on <http://www.ti-lpg.org> "Linux, y'a moins bien mais c'est plus cher !" |
From: Julien B. <jb...@jb...> - 2003-01-05 15:53:24
|
Romain Lievin <ro...@ti...> wrote: >> Do not touch the CVS without asking me first, there's a tilp module >> that in fact resolves to tilp-osx. I need to remove that before. > > I attempted but your cellular phone is not reachable.... Yeah I hit the f*cking wrong key while retrieving the phone :) I'll be unreachable tonight (from now on), and I've got exams this week. (sucks) >> Importing the code for the libs into the CVS will require updating the >> OSX support for them, otherwise we'll have some license issue. > > Really ? Yeah, need to check whether I made source packages for the latest binaries I put up on SF. >> Regarding the read_ico() functions, it is completely unportable. Get >> rid of it ASAP. > > Why are theynon portable ? You're fread()'ing structs in the .ico file. The problem is that memory alignment is not guaranteed to be the same depending on the compiler and the architecture. It works on i386, but it doesn't work on PowerPC. That's one problem, the other being endianness. Remember, that's why I redesigned the skin format a while ago. > Ok, I will convert them... convert is your friend :) JB. --=20 CJ> Les censeurs agitent plus de vent que les moulins des Pays Bas. Tiens, je savais pas que c'=E9taient les moulins qui cr=E9aient le vent. -+- GR in GNU : Dame qui se shoote et sang chaud pensa -+- |
From: Romain L. <ro...@ti...> - 2003-01-05 16:15:57
|
Hi, > Yeah I hit the f*cking wrong key while retrieving the phone :) > > I'll be unreachable tonight (from now on), and I've got exams this > week. (sucks) ok, I will not bother you then. > > >> Importing the code for the libs into the CVS will require updating the > >> OSX support for them, otherwise we'll have some license issue. > > > > Really ? > > Yeah, need to check whether I made source packages for the latest > binaries I put up on SF. Tell me when I will be able to use/modify CVS. In the meanwhile, I will put the TiEmu one... > > >> Regarding the read_ico() functions, it is completely unportable. Get > >> rid of it ASAP. > > > > Why are theynon portable ? > > You're fread()'ing structs in the .ico file. The problem is that > memory alignment is not guaranteed to be the same depending on the > compiler and the architecture. > I see... This is the reason why my bmp loader does not work under LInux ;-) > It works on i386, but it doesn't work on PowerPC. > > That's one problem, the other being endianness. > > Remember, that's why I redesigned the skin format a while ago. > > > Ok, I will convert them... > > convert is your friend :) > I will do it ASAP... > JB. > > -- > CJ> Les censeurs agitent plus de vent que les moulins des Pays Bas. > Tiens, je savais pas que c'?taient les moulins qui cr?aient le vent. > -+- GR in GNU : Dame qui se shoote et sang chaud pensa -+- > > roms. > ------------------------------------------------------- > This sf.net email is sponsored by:ThinkGeek > Welcome to geek heaven. > http://thinkgeek.com/sf > _______________________________________________ > TiLP-devel mailing list > TiL...@li... > https://lists.sourceforge.net/lists/listinfo/tilp-devel -- Romain Lievin, aka 'roms' <ro...@ti...> The TiLP project is on <http://www.ti-lpg.org> "Linux, y'a moins bien mais c'est plus cher !" |
From: Julien B. <jb...@jb...> - 2003-01-05 16:54:46
|
Romain Lievin <ro...@ti...> wrote: >> Yeah, need to check whether I made source packages for the latest >> binaries I put up on SF. > > Tell me when I will be able to use/modify CVS. In the meanwhile, I will= put the > TiEmu one... There's code in this one too ;) the libti68k should be up there, but this doesn't matter as it's deprecated now :) >> You're fread()'ing structs in the .ico file. The problem is that >> memory alignment is not guaranteed to be the same depending on the >> compiler and the architecture. > > I see... This is the reason why my bmp loader does not work under LInux= ;-) Might be :) =20 >> > Ok, I will convert them... >>=20 >> convert is your friend :) > > I will do it ASAP... Check out the icoutils package in Debian unstable, it has a ico -> xpm tool. JB. --=20 Je poste des messages de demeur=E9 parce que j'ai une r=E9putation de demeur=E9 =E0 d=E9fendre. On n'imagine pas combien une telle r=E9putatio= n est difficile =E0 =E9tablir. Son entretien est un souci permanent... -+- EB in Guide du Neuneu sur Usenet : Une r=E9putation, deux meur=E9s. = -+- |
From: Romain L. <ro...@ti...> - 2003-01-05 17:00:58
|
Hi, On Sun, Jan 05, 2003 at 05:54:38PM +0100, Julien BLACHE wrote: > Romain Lievin <ro...@ti...> wrote: > > >> Yeah, need to check whether I made source packages for the latest > >> binaries I put up on SF. > > > > Tell me when I will be able to use/modify CVS. In the meanwhile, I will put the > > TiEmu one... > > There's code in this one too ;) the libti68k should be up there, but > this doesn't matter as it's deprecated now :) > Yes, I have just seen this :-( > >> You're fread()'ing structs in the .ico file. The problem is that > >> memory alignment is not guaranteed to be the same depending on the > >> compiler and the architecture. > > > > I see... This is the reason why my bmp loader does not work under LInux ;-) > > Might be :) > > >> > Ok, I will convert them... > >> > >> convert is your friend :) > > > > I will do it ASAP... > > Check out the icoutils package in Debian unstable, it has a ico -> xpm > tool. BTW, I have upgraded my Debian thru Internet. ADSL is wonderful :-)) > > JB. > > -- > Je poste des messages de demeur? parce que j'ai une r?putation de > demeur? ? d?fendre. On n'imagine pas combien une telle r?putation est > difficile ? ?tablir. Son entretien est un souci permanent... LOL ! > -+- EB in Guide du Neuneu sur Usenet : Une r?putation, deux meur?s. -+- > > > ------------------------------------------------------- > This sf.net email is sponsored by:ThinkGeek > Welcome to geek heaven. > http://thinkgeek.com/sf > _______________________________________________ > TiLP-devel mailing list > TiL...@li... > https://lists.sourceforge.net/lists/listinfo/tilp-devel roms. -- Romain Lievin, aka 'roms' <ro...@ti...> The TiLP project is on <http://www.ti-lpg.org> "Linux, y'a moins bien mais c'est plus cher !" |
From: Romain L. <ro...@ti...> - 2003-01-05 15:36:29
|
Hi, > I also added LIBS="$LIBS $INTLLIBS" in every ./configure. With this Added. > ** libticables 3.5.3 > > I added a few lines to ./configure to define __FBSD__. These should be > added to configure.ac though. I didn't because then I prevented Added. > Further on, the serial devices on FreeBSD are named > /dev/cuaa(0|1|2|3). The parallel ports are /dev/lpt(0|1|2), but I You will have to add some defines in cabl_def.h. > On another note, FreeBSD doesn't define O_SYNC. With O_FSYNC however, > things work perfect. > Linux doesn't have O_FSYNC. > > ** libtifiles 0.4.0 > > ** libticalcs 4.2.8 > > In src/calc_def.h I replaced <glib-1.2/glib.h> with <glib.h>, since I > believe the glib-1.2 part is system dependent and ./configure adds a > proper -I(include) to CFLAGS anyway. The downside is that TiLP's > configure script chokes when checking for <tilp/calc_def.h>, but this > can be overcome by adding the proper -I(include) to CPPFLAGS. I know it's problematic. This problem does not exist any longer with Glib2.0 which is properly detected (pkg-config power !). > ** tilp 6.00 pre15 > > On some lines in the configure script $(prefix) is used. This gave > some problems so I replaced them by $prefix. Should there be any > difference between these two? replaced by ${prefix}, as emphasized by JB. > That should be it. I hope this can be of help to somebody. I'm not > submitting anything to the FreeBSD ports tree yet as I first want to > check a few more things, maybe take a look at the usb lib/driver and > that's all for after exams. At the moment I'm just happy to be able to > connect to my calc! Well, I merged some of yours remarks but not at all. I will prefer a complete patch. If you intend to do this, ask me my latest tree or use the CVS (if I manage to use it...). > > Tijl Coosemans > > _Thanks_, Romain. > > ------------------------------------------------------- > This sf.net email is sponsored by:ThinkGeek > Welcome to geek heaven. > http://thinkgeek.com/sf > _______________________________________________ > TiLP-devel mailing list > TiL...@li... > https://lists.sourceforge.net/lists/listinfo/tilp-devel > -- Romain Lievin, aka 'roms' <ro...@ti...> The TiLP project is on <http://www.ti-lpg.org> "Linux, y'a moins bien mais c'est plus cher !" |
From: Romain L. <ro...@ti...> - 2003-01-10 14:18:09
|
Hi, given that I have ADSL for now, I have installed a server. Developers who need to get the latest source code of my various projects can use my ftp server: ftp://de...@rl... or ftp rlievin.dyndns.org login: devel pass: devel Bandwidth: 128/64 kbits. This server is for developers _only_. Regards, roms. On Sat, Jan 04, 2003 at 11:28:40PM +0100, Tijl Coosemans wrote: > Hello to all TiLP developers, > > Today I finished porting TiLP to FreeBSD. A set of patches for each > package is available at http://tijl.studentenweb.org/tilp/ > Currently I only have the Grey TIGL working. I might take a look at > the others when I finish my exams at the end of January. > > What follows are a few notes about the patch sets. > > > ** General > > FreeBSD currently doesn't have <stdint.h>. I believe the upcoming 5.0 > will support it though. To overcome this I encapsulated each #include > <stdint.h> with: > > +#ifdef HAVE_STDINT_H > #include <stdint.h> > +#else > +#include <inttypes.h> > +#endif > > Most of the patches just do this. I realise this isn't really a nice > structure to have in your code, so you may not want to apply these. > > I also added LIBS="$LIBS $INTLLIBS" in every ./configure. With this > line each package can be compiled with gettext in libintl in case glib12 > isn't available. For libticalcs and tilp this doesn't matter much since > they depend on glib12 anyway. > > Next I noticed the ti*-config scripts had a bug when the --cflags > option was given. They returned the dir without the -I prefix. > > > ** libticables 3.5.3 > > I added a few lines to ./configure to define __FBSD__. These should be > added to configure.ac though. I didn't because then I prevented > using autoconf/automake/... during compilation. I wonder why you don't > use any gcc predefined constants though. I believe either unix or > __unix__ is defined on every unix platform. This is what I get on a > FreeBSD box. > > tijl tijl% gcc -dM -E - < /dev/null > #define __FreeBSD__ 4 > #define __FreeBSD_cc_version 460001 > #define __i386__ 1 > #define __i386 1 > #define __GNUC_MINOR__ 95 > #define i386 1 > #define __unix 1 > #define __unix__ 1 > #define __GNUC__ 2 > #define __ELF__ 1 > #define unix 1 > tijl tijl% > > Further on, the serial devices on FreeBSD are named > /dev/cuaa(0|1|2|3). The parallel ports are /dev/lpt(0|1|2), but I > haven't added these yet, since the parallel cable isn't supported yet > anyway. > > On another note, FreeBSD doesn't define O_SYNC. With O_FSYNC however, > things work perfect. > > > ** libtifiles 0.4.0 > > No system specific patches were necessary. <stdint.h> is used quite a > lot though. > > ** libticalcs 4.2.8 > > In src/calc_def.h I replaced <glib-1.2/glib.h> with <glib.h>, since I > believe the glib-1.2 part is system dependent and ./configure adds a > proper -I(include) to CFLAGS anyway. The downside is that TiLP's > configure script chokes when checking for <tilp/calc_def.h>, but this > can be overcome by adding the proper -I(include) to CPPFLAGS. > At the moment this is done automatically by a Makefile I made for > TiLP's entry in the FreeBSD ports tree, but I can imagine it isn't so > convenient when you have to run ./configure manually. Probably, some > construction can be figured out, like adding a CPPFLAGS= line at the > same place as LIBS= CFLAGS= lines currently in there somewhere. > Related to this, I could comment out a redefinition of > HAVE_TILP_TICALCS_H in TiLP's src/tilibs.h. > > > ** tilp 6.00 pre15 > > On some lines in the configure script $(prefix) is used. This gave > some problems so I replaced them by $prefix. Should there be any > difference between these two? > > src/term_cb.c wants to include error.h, but there doesn't seem to be > any. > > > ** > > That should be it. I hope this can be of help to somebody. I'm not > submitting anything to the FreeBSD ports tree yet as I first want to > check a few more things, maybe take a look at the usb lib/driver and > that's all for after exams. At the moment I'm just happy to be able to > connect to my calc! > > Tijl Coosemans > > > > ------------------------------------------------------- > This sf.net email is sponsored by:ThinkGeek > Welcome to geek heaven. > http://thinkgeek.com/sf > _______________________________________________ > TiLP-devel mailing list > TiL...@li... > https://lists.sourceforge.net/lists/listinfo/tilp-devel > -- Romain Lievin, aka 'roms' <ro...@ti...> The TiLP project is on <http://www.ti-lpg.org> "Linux, y'a moins bien mais c'est plus cher !" |