Re: [Fish-users] seq cannot count backwards
On 04/02/10 01:31, David Frascone wrote: Found and fixed. There were several issues. First, most people who type seq are really running seq on their host. Fish will only use the builtin if it doesn't find it locally. Use 'seq --version' to see what I mean. of course seq is /usr/bin/seq ! (or wherever it is on your path.) What does it have to do with Fish? How can Fish have a possibly-a-builtin, possibly-not?(for me, 'type seq' just says 'seq is /usr/bin/seq' ...) Isn't it against Fish's philosophy to duplicate external tools that don't need to be built into a shell? Is Mac OS X 'seq' broken, under-featured, (or nonexistent?)? I would be unsurprised. In 10.3 (the last version I used regularly), I know they shipped a version of 'find' that enjoyed segfaulting (or some weird error, I forget exactly) when you forgot that their version of the 'find' command didn't support omitting the path bit (you had to pass '.'). Admittedly, I think they just copied the tools from BSD, but that doesn't mean they were good tools... -Isaac -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users
Re: [Fish-users] seq cannot count backwards
On 2 Apr 2010, at 10:52, Michael Lachmann wrote: I think when fish is invoked for a script, it shouldn't parse the arguments that are meant for the script... I think one solution is to add a '+' to the beginning of the options string in fish.c: #define GETOPT_STRING +hilnvc:p:d: That will stop options processing once the script name is hit. Michael On 2 Apr 2010, at 8:09, Isaac Dupree wrote: On 04/02/10 01:31, David Frascone wrote: Found and fixed. There were several issues. First, most people who type seq are really running seq on their host. Fish will only use the builtin if it doesn't find it locally. Use 'seq --version' to see what I mean. of course seq is /usr/bin/seq ! (or wherever it is on your path.) What does it have to do with Fish? How can Fish have a possibly-a-builtin, possibly-not?(for me, 'type seq' just says 'seq is /usr/bin/seq' ...) Isn't it against Fish's philosophy to duplicate external tools that don't need to be built into a shell? Is Mac OS X 'seq' broken, under-featured, (or nonexistent?)? I would be unsurprised. In 10.3 (the last version I used regularly), I know they shipped a version of 'find' that enjoyed segfaulting (or some weird error, I forget exactly) when you forgot that their version of the 'find' command didn't support omitting the path bit (you had to pass '.'). Admittedly, I think they just copied the tools from BSD, but that doesn't mean they were good tools... -Isaac -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users
Re: [Fish-users] seq cannot count backwards
Right. I was in the same boat. Use my patch and the fish installed seq will work for you. -Dave Sent from my iPhone On Apr 2, 2010, at 4:52 AM, Michael Lachmann lachm...@eva.mpg.de wrote: The seq that I use is /sw/bin/seq, which seems to have been installed by fish: --- #!/usr/bin/env fish # # Fallback implementation of the seq command # # seq. Generated from seq.in by configure. set -l from 1 . . . --- The error (fish: invalid option -- 1) seems to be generated before the script is ever called, by fish itself. So, when the script is invoked, fish is called, with the arguments (10 -1 5), and it generates the error. I think when fish is invoked for a script, it shouldn't parse the arguments that are meant for the script... Michael On 2 Apr 2010, at 8:09, Isaac Dupree wrote: On 04/02/10 01:31, David Frascone wrote: Found and fixed. There were several issues. First, most people who type seq are really running seq on their host. Fish will only use the builtin if it doesn't find it locally. Use 'seq --version' to see what I mean. of course seq is /usr/bin/seq ! (or wherever it is on your path.) What does it have to do with Fish? How can Fish have a possibly-a-builtin, possibly-not?(for me, 'type seq' just says 'seq is /usr/bin/seq' ...) Isn't it against Fish's philosophy to duplicate external tools that don't need to be built into a shell? Is Mac OS X 'seq' broken, under-featured, (or nonexistent?)? I would be unsurprised. In 10.3 (the last version I used regularly), I know they shipped a version of 'find' that enjoyed segfaulting (or some weird error, I forget exactly) when you forgot that their version of the 'find' command didn't support omitting the path bit (you had to pass '.'). Admittedly, I think they just copied the tools from BSD, but that doesn't mean they were good tools... -Isaac --- --- --- - Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users --- --- --- - Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users
Re: [Fish-users] seq cannot count backwards
Seq does not exist on OS X, it seems. Personally I've never used it. The patch only fixes the version installed by fish, so, should have no side effects at all! -Dave Sent from my iPhone On Apr 2, 2010, at 2:09 AM, Isaac Dupree m...@isaac.cedarswampstudios.org wrote: On 04/02/10 01:31, David Frascone wrote: Found and fixed. There were several issues. First, most people who type seq are really running seq on their host. Fish will only use the builtin if it doesn't find it locally. Use 'seq --version' to see what I mean. of course seq is /usr/bin/seq ! (or wherever it is on your path.) What does it have to do with Fish? How can Fish have a possibly-a-builtin, possibly-not?(for me, 'type seq' just says 'seq is /usr/bin/seq' ...) Isn't it against Fish's philosophy to duplicate external tools that don't need to be built into a shell? Is Mac OS X 'seq' broken, under-featured, (or nonexistent?)? I would be unsurprised. In 10.3 (the last version I used regularly), I know they shipped a version of 'find' that enjoyed segfaulting (or some weird error, I forget exactly) when you forgot that their version of the 'find' command didn't support omitting the path bit (you had to pass '.'). Admittedly, I think they just copied the tools from BSD, but that doesn't mean they were good tools... -Isaac --- --- --- - Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users
Re: [Fish-users] seq cannot count backwards
What OS are you using? Also, Ben has not pulled my changes yet, so, you can either apply my patches by hand, or pull from my tree: git clone git://github.com/CodeMonk/fish.git -Dave On Fri, Apr 2, 2010 at 7:07 AM, Michael Lachmann lachm...@eva.mpg.dewrote: I tried to download using git, but I'm getting an error, and I'm not sure what I did wrong. I did this: git clone git://github.com/benhoskings/fish cd fish ./configure make sudo make install but, calling fish gives me an error: ~/D/f/fish /usr/local/bin/fish fish: Job 1, '/usr/local/bin/fish' terminated by signal SIGBUS (Misaligned address error) ? On 2 Apr 2010, at 13:03, David Frascone wrote: Right. I was in the same boat. Use my patch and the fish installed seq will work for you. -Dave Sent from my iPhone On Apr 2, 2010, at 4:52 AM, Michael Lachmann lachm...@eva.mpg.de wrote: The seq that I use is /sw/bin/seq, which seems to have been installed by fish: --- #!/usr/bin/env fish # # Fallback implementation of the seq command # # seq. Generated from seq.in by configure. set -l from 1 . . . --- The error (fish: invalid option -- 1) seems to be generated before the script is ever called, by fish itself. So, when the script is invoked, fish is called, with the arguments (10 -1 5), and it generates the error. I think when fish is invoked for a script, it shouldn't parse the arguments that are meant for the script... Michael On 2 Apr 2010, at 8:09, Isaac Dupree wrote: On 04/02/10 01:31, David Frascone wrote: Found and fixed. There were several issues. First, most people who type seq are really running seq on their host. Fish will only use the builtin if it doesn't find it locally. Use 'seq --version' to see what I mean. of course seq is /usr/bin/seq ! (or wherever it is on your path.) What does it have to do with Fish? How can Fish have a possibly-a- builtin, possibly-not?(for me, 'type seq' just says 'seq is /usr/bin/ seq' ...) Isn't it against Fish's philosophy to duplicate external tools that don't need to be built into a shell? Is Mac OS X 'seq' broken, under-featured, (or nonexistent?)? I would be unsurprised. In 10.3 (the last version I used regularly), I know they shipped a version of 'find' that enjoyed segfaulting (or some weird error, I forget exactly) when you forgot that their version of the 'find' command didn't support omitting the path bit (you had to pass '.'). Admittedly, I think they just copied the tools from BSD, but that doesn't mean they were good tools... -Isaac -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users
Re: [Fish-users] seq cannot count backwards
Thanks! I got your version now. git seems pretty nice. Anyway, the problem was in config.h - I have to comment out the HAVE_REALPATH_NULL: include/config.h:/*#define HAVE_REALPATH_NULL 1*/ I'm using OSX 10.5.8, I don't know why my realpath can't handle NULL. I think two additional fixes are needed: 1. in kill.c, add the option '-n' to echo, so that fish doesn't add an extra newline to what's copied to the buffers. 2. in fish.c add + to the getopt: #define GETOPT_STRING +hilnvc:p:d: That tells fish to stop reading arguments when it encounters a non- option. That caused a bug in seq, but it'll cause a bug in any fish script that has to accept options or negative numbers. Michael On 2 Apr 2010, at 15:53, David Frascone wrote: What OS are you using? Also, Ben has not pulled my changes yet, so, you can either apply my patches by hand, or pull from my tree: git clone git://github.com/CodeMonk/fish.git -Dave On Fri, Apr 2, 2010 at 7:07 AM, Michael Lachmann lachm...@eva.mpg.de wrote: I tried to download using git, but I'm getting an error, and I'm not sure what I did wrong. I did this: git clone git://github.com/benhoskings/fish cd fish ./configure make sudo make install but, calling fish gives me an error: ~/D/f/fish /usr/local/bin/fish fish: Job 1, '/usr/local/bin/fish' terminated by signal SIGBUS (Misaligned address error) ? On 2 Apr 2010, at 13:03, David Frascone wrote: Right. I was in the same boat. Use my patch and the fish installed seq will work for you. -Dave Sent from my iPhone On Apr 2, 2010, at 4:52 AM, Michael Lachmann lachm...@eva.mpg.de wrote: The seq that I use is /sw/bin/seq, which seems to have been installed by fish: --- #!/usr/bin/env fish # # Fallback implementation of the seq command # # seq. Generated from seq.in by configure. set -l from 1 . . . --- The error (fish: invalid option -- 1) seems to be generated before the script is ever called, by fish itself. So, when the script is invoked, fish is called, with the arguments (10 -1 5), and it generates the error. I think when fish is invoked for a script, it shouldn't parse the arguments that are meant for the script... Michael On 2 Apr 2010, at 8:09, Isaac Dupree wrote: On 04/02/10 01:31, David Frascone wrote: Found and fixed. There were several issues. First, most people who type seq are really running seq on their host. Fish will only use the builtin if it doesn't find it locally. Use 'seq --version' to see what I mean. of course seq is /usr/bin/seq ! (or wherever it is on your path.) What does it have to do with Fish? How can Fish have a possibly-a- builtin, possibly-not?(for me, 'type seq' just says 'seq is /usr/bin/ seq' ...) Isn't it against Fish's philosophy to duplicate external tools that don't need to be built into a shell? Is Mac OS X 'seq' broken, under-featured, (or nonexistent?)? I would be unsurprised. In 10.3 (the last version I used regularly), I know they shipped a version of 'find' that enjoyed segfaulting (or some weird error, I forget exactly) when you forgot that their version of the 'find' command didn't support omitting the path bit (you had to pass '.'). Admittedly, I think they just copied the tools from BSD, but that doesn't mean they were good tools... -Isaac -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Fish-users mailing list Fish-users@lists.sourceforge.net
Re: [Fish-users] seq cannot count backwards
On Fri, Apr 2, 2010 at 10:06 AM, Michael Lachmann lachm...@eva.mpg.dewrote: Thanks! I got your version now. git seems pretty nice. Anyway, the problem was in config.h - I have to comment out the HAVE_REALPATH_NULL: include/config.h:/*#define HAVE_REALPATH_NULL 1*/ I'm using OSX 10.5.8, I don't know why my realpath can't handle NULL. I think two additional fixes are needed: 1. in kill.c, add the option '-n' to echo, so that fish doesn't add an extra newline to what's copied to the buffers. Doh! I wondered what was causing that -- my bad. I'll fix push it. 2. in fish.c add + to the getopt: #define GETOPT_STRING +hilnvc:p:d: That tells fish to stop reading arguments when it encounters a non-option. That caused a bug in seq, but it'll cause a bug in any fish script that has to accept options or negative numbers. Will fix that too. (fixing now) -- done. You can do a git pull now. -Dave Michael On 2 Apr 2010, at 15:53, David Frascone wrote: What OS are you using? Also, Ben has not pulled my changes yet, so, you can either apply my patches by hand, or pull from my tree: git clone git://github.com/CodeMonk/fish.git -Dave On Fri, Apr 2, 2010 at 7:07 AM, Michael Lachmann lachm...@eva.mpg.de wrote: I tried to download using git, but I'm getting an error, and I'm not sure what I did wrong. I did this: git clone git://github.com/benhoskings/fish cd fish ./configure make sudo make install but, calling fish gives me an error: ~/D/f/fish /usr/local/bin/fish fish: Job 1, '/usr/local/bin/fish' terminated by signal SIGBUS (Misaligned address error) ? On 2 Apr 2010, at 13:03, David Frascone wrote: Right. I was in the same boat. Use my patch and the fish installed seq will work for you. -Dave Sent from my iPhone On Apr 2, 2010, at 4:52 AM, Michael Lachmann lachm...@eva.mpg.de wrote: The seq that I use is /sw/bin/seq, which seems to have been installed by fish: --- #!/usr/bin/env fish # # Fallback implementation of the seq command # # seq. Generated from seq.in by configure. set -l from 1 . . . --- The error (fish: invalid option -- 1) seems to be generated before the script is ever called, by fish itself. So, when the script is invoked, fish is called, with the arguments (10 -1 5), and it generates the error. I think when fish is invoked for a script, it shouldn't parse the arguments that are meant for the script... Michael On 2 Apr 2010, at 8:09, Isaac Dupree wrote: On 04/02/10 01:31, David Frascone wrote: Found and fixed. There were several issues. First, most people who type seq are really running seq on their host. Fish will only use the builtin if it doesn't find it locally. Use 'seq --version' to see what I mean. of course seq is /usr/bin/seq ! (or wherever it is on your path.) What does it have to do with Fish? How can Fish have a possibly-a- builtin, possibly-not?(for me, 'type seq' just says 'seq is /usr/bin/ seq' ...) Isn't it against Fish's philosophy to duplicate external tools that don't need to be built into a shell? Is Mac OS X 'seq' broken, under-featured, (or nonexistent?)? I would be unsurprised. In 10.3 (the last version I used regularly), I know they shipped a version of 'find' that enjoyed segfaulting (or some weird error, I forget exactly) when you forgot that their version of the 'find' command didn't support omitting the path bit (you had to pass '.'). Admittedly, I think they just copied the tools from BSD, but that doesn't mean they were good tools... -Isaac -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got
Re: [Fish-users] Fixing title setting in OS X
On 04/02/10 10:26, David Frascone wrote: So, how do you think this should be generically fixed? In my tree, I simply commented out the /dev/tty setting. But, I don't want to break things for linux. And, I don't want to have conditional compilations, if they can be avoided. as you wish... I think conditional compilation for accessing Linux vs. Darwin /dev/ filenames makes a ton of sense. (Or use a C if() and a system-call that tells you which system it is, if you wish, if such a thing exists... it's conventional to use conditional-compilation!) But you'll have to tell me, what is the exact naming convention that Darwin/OSX uses? (for example, is /dev/tty1 a device-name on Darwin, or not? If not, then we can tell the difference just based on the names, with just a small bit of coding.) Should I abuse another environment variable? Even still, how should it look? #linux set FISH_TERMINALS_WITH_TITLES /dev/pts whatever screen uses #osx set FISH_TERMINALS_WITH_TITLES /dev/tty whatever screen uses Then the code can go through that and do the strnstr? I don't particularly like adding more special environment variables either, but I like it better than conditional compilation. I suppose special fish environment variables could be used... make sure that they're initialized correctly for each system though, even if you share ~/.config/fish/ between a Mac and a Linux system and go back and forth... and it would clearly be just as much conditional as compiling it into the C code. And I can tell that you're not engineering with weird hackers on weird embedded Linux setups in mind, so this configurability would probably not be useful even for dealing with that. -Isaac -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users
Re: [Fish-users] Fixing title setting in OS X
On Fri, Apr 2, 2010 at 1:23 PM, Isaac Dupree m...@isaac.cedarswampstudios.org wrote: as you wish... I think conditional compilation for accessing Linux vs. Darwin /dev/ filenames makes a ton of sense. (Or use a C if() and a system-call that tells you which system it is, if you wish, if such a thing exists... it's conventional to use conditional-compilation!) But you'll have to tell me, what is the exact naming convention that Darwin/OSX uses? (for example, is /dev/tty1 a device-name on Darwin, or not? If not, then we can tell the difference just based on the names, with just a small bit of coding.) Good point: Darwin is using: codemon...@daves-mbp ~/dotfiles tty /dev/ttys015 I think ttys may be safe to use! I suppose special fish environment variables could be used... make sure that they're initialized correctly for each system though, even if you share ~/.config/fish/ between a Mac and a Linux system and go back and forth... and it would clearly be just as much conditional as compiling it into the C code. And I can tell that you're not engineering with weird hackers on weird embedded Linux setups in mind, so this configurability would probably not be useful even for dealing with that. So, given the ttys difference -- think we should go with an env setting, or a quick fix that works for Darwin? I need to get some virtual machines set up anyway, I'll try to set up a BSD one too, to see if that breaks. . . -Dave -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users
Re: [Fish-users] Fixing title setting in OS X
On 04/02/10 13:34, David Frascone wrote: So, given the ttys difference -- think we should go with an env setting, or a quick fix that works for Darwin? just use the appropriate check for ttys in the code. And make sure that you add a comment for what each string-check is checking for (e.g. Darwin terminals, Linux ptys, Linux console terminals... it's a dark area of Unix, so just be clear enough that people can go search the internet to learn more if they need to fix it or something). -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users
Re: [Fish-users] Fixing title setting in OS X
On Fri, Apr 2, 2010 at 2:17 PM, Isaac Dupree m...@isaac.cedarswampstudios.org wrote: On 04/02/10 13:34, David Frascone wrote: So, given the ttys difference -- think we should go with an env setting, or a quick fix that works for Darwin? just use the appropriate check for ttys in the code. And make sure that you add a comment for what each string-check is checking for (e.g. Darwin terminals, Linux ptys, Linux console terminals... it's a dark area of Unix, so just be clear enough that people can go search the internet to learn more if they need to fix it or something). Ok -- code done. Seems to work well on OS X. http://github.com/CodeMonk/fish/commit/e11e78dfd762b80ccd4172a740e38fee0dc2c1e0 I need to check out a linux tree from darcs and apply these changes to make sure I'm not breaking anything . . . looks much cleaner though. Since tty is a substring of ttys, there had to be a nested check, but, seems pretty obvious by the comments. (If not, let me know, and I'll be even more verbose) -Dave -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users
Re: [Fish-users] Temporary functions?
On Fri, Apr 2, 2010 at 6:07 PM, Philip Ganchev phil.ganc...@gmail.comwrote: On Thu, Apr 1, 2010 at 10:08 AM, David Frascone d...@frascone.com wrote: I'd like to add a function to my config.fish to do some things over and over [...] But, I don't want this function to exist after config.fish is done. Should I just erase it before the end of my config.fish: functions -e add_to_path Or is there an easier way to make a function that is non-global in scope? I think you should just erase it. There is no way to define a non-global function, as far as I know. Is there a better way than a function to do what I want? (Uniquely add directories to my path)? -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users
Re: [Fish-users] Uniquely add to path?
On Thu, Apr 1, 2010 at 9:46 AM, David Frascone d...@frascone.com wrote: There's a really cool feature in zsh that allows you to uniquely add a string to the path, if it did not exist before: typeset -U path This makes all items in path have to be unique, so then, if you do: path=(/foo/bin $path) path=(/foot/bin $path) You will NOT have two /foo/bin(s) in your path. Is there an equivalent way to do this in fish? Not that I know of. Fish is designed to have a minimal set of builtins that are as orthogonal as possible. But it might be useful to define a function like: function varuniq --description Return the unique arguments, merging to the first occurrence of a word. for x in $argv; if not contains $x $ret set ret $ret $y end end end Then use it like this: set PATH (varuniq $PATH /home/joe/bin home/joe/src) Or define: function typeset set rest argv[(seq 2 (count $argv))] set $argv[1] (varuniq $argv[1] $rest) end Or even: function path for x in argv if test -f $x and not contains $PATH $x set PATH $PATH $x end end end -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users
Re: [Fish-users] Uniquely add to path?
I like the third function . . .. but I'd do a -d $x, since it should be a directory :) -Dave On Fri, Apr 2, 2010 at 7:15 PM, Philip Ganchev phil.ganc...@gmail.comwrote: On Thu, Apr 1, 2010 at 9:46 AM, David Frascone d...@frascone.com wrote: There's a really cool feature in zsh that allows you to uniquely add a string to the path, if it did not exist before: typeset -U path This makes all items in path have to be unique, so then, if you do: path=(/foo/bin $path) path=(/foot/bin $path) You will NOT have two /foo/bin(s) in your path. Is there an equivalent way to do this in fish? Not that I know of. Fish is designed to have a minimal set of builtins that are as orthogonal as possible. But it might be useful to define a function like: function varuniq --description Return the unique arguments, merging to the first occurrence of a word. for x in $argv; if not contains $x $ret set ret $ret $y end end end Then use it like this: set PATH (varuniq $PATH /home/joe/bin home/joe/src) Or define: function typeset set rest argv[(seq 2 (count $argv))] set $argv[1] (varuniq $argv[1] $rest) end Or even: function path for x in argv if test -f $x and not contains $PATH $x set PATH $PATH $x end end end -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev___ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users
Re: [Fish-users] seq cannot count backwards
On 04/02/10 04:52, Michael Lachmann wrote:
The seq that I use is /sw/bin/seq, which seems to have been installed
by fish:
---
#!/usr/bin/env fish
#
# Fallback implementation of the seq command
gee.
Are there, I wonder, any systems with Fish installed but not Bash (or at
least some bourne-compatible/posix shell as /bin/sh)?
Yes, I admit I do my scripting in Bash. Not perl, not python, not zsh,
not oh-er-there-must-be-another-scripting-language. Bash is everywhere,
bash is reliable ('cause everyone uses it, so everyone notices its
bugs), and you basically have to know bash if you're to accurately read
the advice and scripts that people give on the Internet*, let alone to
give advice. (*maybe you don't need to know the looping constructs,
history substitution, keyboard shortcuts.. -- but then again, maybe...)
The error (fish: invalid option -- 1) seems to be generated before
the script is ever called, by fish itself.
So, when the script is invoked, fish is called, with the arguments (10
-1 5), and it generates the error.
I think when fish is invoked for a script, it shouldn't parse the
arguments that are meant for the script...
Yes, that would be a Fish bug, providing that Fish wishes to support the
existence of Fish-scripts at all. (I guess we would be silly not to
technologically support it.. regardless of whether/which people say it's
a good idea for common use, it shouldn't be broken!)
-Isaac
--
Download Intel#174; Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
___
Fish-users mailing list
Fish-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/fish-users
