Hi,
The patchset again with small differencies and a few additions.
Regards,
>From 15608429bf80283f2edaf8bed04fb71e986b0123 Mon Sep 17 00:00:00 2001
From: "Roberto E. Vargas Caballero"
Date: Sat, 11 Apr 2020 11:52:58 +0200
Subject: [PATCH 5/6] Update FAQ with the last odifications
---
FAQ |
Hi,
As the original author of ed I have to apologize about this kind
of errors. I didn;t have so much time to finalize it, and there is
a list of known bugs in the header of the file. I would recomend to
add new known bugs to the the list and them let to new people to
fix them. I think the code
On Sat, Dec 21, 2019 at 05:00:53AM -0300, Wilson M. S. Moncayo wrote:
> The tmux terminfo extensions Ss and Se are respectively used to switch
> and reset the cursor style, and were finally made to work in st in
> commit 8386642, wherein the string "\E[2 q" is used to reset the cursor
> style.
> So for lexing, usually a simple token parser in C will do the job
> better. And for parsing, you get the problem that yacc will create an
> LALR parser, which is a bottom-up parser. Which may be faster but
> doesn't allow for good error messages on faulty input ("error: expected
> this, that, or
> Not all keyboards have a logo key, most of them have an alt key though.
My keyboard doesn't have a win key.
>> "My 2c": I would prefer shell "printf" than "echo -n -e"
>
> yeah, good point. Any of which works.
Yes, but echo -n is not POSIX.
> Oops, I misread the diff. Agreed.
Well, my idea wasn't accept the patch (I hate the ifdef there), but
changing the clock used and remove the problem.
Regards,
> Update your system.
I think the important point here is if it is POSIX. If it is not
then maybe we should think to revert to CLOCK_MONOTONIC.
Regards,
> > The only issue I see with c99 code is that some of the compilers
> > appear to be behind the times on c. Any reason why we wouldn???t want
> > to use a newer c feature other than compatibility?
>
> it's always about weighing convenience against freedom. Modern language
> standards have picked
> I you pointed to a perfect example. While extremely it is extremely
> convenient for some architectures to represent strings as a pointer to
> char/bytes + an implicit terminator, Pascal strings (really, a
> struct/object containing pointer + length) are imminently safer. As
This is not true.
> This way, we can call it easily in the code without troubles.
> I admittedly am a huge supporter of pledge() and would be happy
Then send the patch to the maintainer of the OpenBSD port.
Regards,
> Terminals accept arbitrary input, so pledge support could be a very desirable
> and well-suited feature.
Then add the lines yourself.
Regards,
> there. The only two event types that get through to this code
> are PropertyNotify and SelectionNotify and for both of them property
> should be deleted.
Indeed. If Christoph has not problems with it, I will push this commit.
Regards,
Hi Markus,
> This change allows st to correctly render fonts given in point sizes,
> bringing its behavior in line with other software: dwm, dmenu, tabbed,
> etc:
>
> FC_DEBUG=1 st -f Terminus:size=10
I think __20h__ should take a look to this patch, he is the expert
in fonts (thankfully for me
> What would be the right way to do something like this?
Using the terminfo database.
> clear(1) from ncurses also clears the scrollback buffer if the terminal
> supports it (see the manpage), that's not what you'd want here. As for
> portability, tput clear does `^[[2J^[[H` for basically everything:
Dimitris, if I send patches to remove these hard coded sequences by
calls to tput
> Is it appropriate to use ANSI escape codes in the program rather than
> using something like tput or terminfo, or to just execute the "clear"
> command? Are the escape codes portable?
No, they are not. The problem here is that a lot of ppl in suckless
think that doesn't matter to keep the
> but simplicity is more important. And yet if ye don't want to use GNU
> syntax for inline assembly, then people would have to
> write code to comply with scc rules.
I think this force them to write these parts in assembler directly,
and it opens the door to the few cases where it can be needed.
> .S or .asm actually, .as is ActionScript. If my two cents are counted,
I know several toolchains that use .as for assembler file.
Regards,
> inline assembly is irrelevant and not needed.
This is my way of thinking, because if you need asm, then your code is
not portable at all, and then it is better to move the asm to a
function and coded it directly in asm in a .as file. I am open to
discuss it, because I can understand a few
> There is no such thing as GNU syntax. gas supports both syntaxes;
> AT is the de facto standard.
I meant the GNU syntax for inline assembly, not the syntax of
the assembler itself. Something that I could accept is for example:
__asm("...");
or
__asm {
...
> After I quit using shift-alt-q it's completely frozen, can't even switch
> between consoles.
Strange. As far as I know, it only can happen if the kernel itself
frozen, or the tty is in raw mode. I suppose the kernel is not
frozen, (we don't code so good code to be able to hang the kernel),
> I forgot to ask, will scc be intended for
> building something big, say, GTK+ programmes,
> or maybe even the kernel? And what about
> inline assembly and compiling assembly code
> itself?
We don't want to follow all the extensions of GNU, this is for sure.
If you try to be compatible with GCC
> I tried to ran cc2 from the directory, but it still hangs.
> Even if I simply try "./cc2". The same applies to
> cc2-amd64-sysv and cc2-i386-sysv.
yes, it is logical because cc2 reads the input data from stdin, so if
you execute it without piping into it the output of cc1 then it will
wait
> I think compilers are way over my level, but it's a subject that interest me
> and
> clang is massive c++, gcc is nonstandard bloat, tcc doesnt seem sufficient,
> so I wanna help.
There are a lot of places were helping without having a big idea about
compilers. The scc driver is one of the
Hi,
> I usually compile suckless software with tcc,
> but it fails at static linking, so I found myself
> cloning scc.
Scc is still a work in progress. It is far of being
completed and being able to compile programs.
> I changed PREFIX in config.mk to /usr/local and
> successfully installed it.
> It's likely that tmux checks some specific terminfo properties
> before relaying 1004. If that's the case, we could temporarily
> override them inside .tmux.conf to confirm and then add to compiled
> st terminfo permanently.
Uhmmm, I dunno. As far as I know there is no terminfo
capability
Hi again,
> We can try to test if we are receiving the event or not.
> The posibilities here are:
>
> - Mode is NotifyGrab
> - We do not receive the event
And the answer was: Any of them. We are not receiving the 1004
set mode sequence. I don't know why, but tmux think that is
Hi,
> I'm glad someone finally tried it.:)
I'm sorry. I am setting my network now, and I cannot read
the mail as fast as I used to do it. BTW, the another
maintainer shold have to answer, but he is a bit lazy ...
> However, I wonder how you could receive this focus-in event:
I mistook, I
Hi again,
> Test case for bug:
> Open empty terminal
> st -e tmux
> Activate focusing
> printf "\e[?1004h"
> Catch one focus event
>
> Move focus to another window
> Expected: printed ^[[O in place of pending ^
> Actual result: nothing changed
I mistook in my previos
Hi,
> Move focus to another window
> Expected: printed ^[[O in place of pending ^
> Actual result: nothing changed
I have tested it in dwm and git st and I receive a ^[[O.
Can you try it with dwm and send the results?
Regards,
> Yeah, I'd really like to get rid of boost, and possibly migrate to a
> lighter regex lib. The problem is that no currently available libs match
> the required feature set:
>
Write it. it is not so somplex to write a regex library.
Regards,
>> Idea: modify `st` so it can change `cursorahape` dynamically, modify
>> `vis` to change the cursor on mode change.
> Maybe I'm crazy, but doesn't st already do this? Neovim is able to change the
> cursor when running in it.
There is a sequence to chnge the cursor shape, and I guess neovim
is
Applied, thanks.
I know this isn't a democracy, but I agree with Greg, it makes more
sense to only set PRIMARY, not CLIPBOARD, in selcopy. Removing the
clipboard-related lines from xsetsel seems to do the trick. I've
attached a patch that does just that.
I agree here, it shouldn't modiy the CLIPBOARD
Hi,
char *name; /* string representation of op */
- int type; /* from Tok.type */
- int prec; /* precedence*/
- int nargs; /* number of arguments (unary or binary) */
- int lassoc; /* left
Hello list,
I know this isn't a democracy, but I agree with Greg, it makes more
sense to only set PRIMARY, not CLIPBOARD, in selcopy. Removing the
clipboard-related lines from xsetsel seems to do the trick. I've
attached a patch that does just that.
I agree here, it shouldn't modiy the
a new version of the patch? Also, I think such a patch should be
considered for inclusion at http://st.suckless.org/patches/. Thank you!
There you go!
Please let me know if it works. Apply with git am.
I submitted a version of the patch to the wiki, and I supposed it already
was there,
I will apply it. Thanks
Regards,
Hi,
your line driver. I mean, st ignores xon/xoff characters, so the line
driver (configured via stty with the flags ixon, ixoff, ixany) is
making the job of stopping the comunication. If we add this patch to
st we must add some information about it in the FAQ (similar to the
information
As per discussion on IRC with FRIGN, ioctl is no standard, but it is
probably a defacto standard; how else can you determine the terminal
width portably? However, this patch adds a fallback in case you can't
read the terminal width - this is how cols currently does it.
You can trust the
Maybe still add an assert() against the divisor != 0 before the code in the
other patch.
Just so people in the future will know that if it does == 0 there is a logic
error elsewhere in the program.
I don't like this idea, because it means we have to add an assert before
any division,
Use the terminfo delay syntax ($x) in our flash capability to avoid
hardcoding a fixed delay in redraw() when called from tsetmode() with
DECSCNM.
We need to turn on the npc capability so that delays are made with
xon/xoff instead of padding characters.
I tried long time ago something
By the way, I looked at other term definitions of this capability and I
noticed this interesting difference:
vt200|vt220, similar for xterm, konsole, etc.
flash=\E[?5h$200/\E[?5l
vt330|vt340, vt400, linux-basic…
flash=\E[?5h\E[?5l$200/
Does
Hi,
This is very inconsistent, because the first linebreak uses the backslash and
the second one directly following it does not use it. I think the first one
should be removed.
It is true that it is inconsistent (and I think it's my fault). I personally
don't like these unneeded \, and I
Firstly, I did not know C required backslashed newlines for
continuation,
It doesn't need it. Backslash escapes newlines, but in this
case is not needed because two strings together are joined.
They are put only because it's part of the style used by
the original author of st.
but a bit more
The ioctl() can be assumed to be present on all systems, but in case
TIOCGWINSZ doesn't exist, we fall back to the fixed value. This is the
best of both worlds.
I like your solution, but I think COLUMNS should have a bigger
priority, because in this case user can define a different value of
On Sun, 15 Feb 2015 19:35:51 +0100
k...@shike2.com wrote:
I like your solution, but I think COLUMNS should have a bigger
priority, because in this case user can define a different value of
the current width.
Okay, then do 3 fallbacks
ioctl - getenv - fixed width
Ups, I didn't
On Tue, 10 Feb 2015 21:21:56 +0100
k...@shike2.com wrote:
Hey k0ga,
When you write a library one of the objects is to minimize
what code is included in the binary when the library is linked.
When you link with a library you extract object files from the
archive, so all the functions
On Tue, Feb 10, 2015 at 10:07:52PM +, Connor Lane Smith wrote:
int
utftorunestr(const char *s, Rune *p)
{
int i, n;
for(i = 0; (n = chartorune(p[i], s)), p[i] != 0; i++)
s += n;
return i;
}
int
fgetrune(Rune *p, FILE *fp)
{
char buf[UTFmax];
Thanks for the feedback everyone. Glad to contribute, st is literally the only
sane freely available graphical terminal emulator in existence.
The original code still looks a bit fishy to me. It looks like there could be
a situation where two newlines are written when a single character
Thanks k0ga. I have not taken the time to check whether this is on the
wiki or on git, but last week I updated my suckless software packages
and patches and to have this was definitely a good thing.
I have uploaded it to the wiki, but I have to add a README about it.
Regards,
On Fri, Jan 30, 2015 at 12:41 AM, FRIGN d...@frign.de wrote:
On Thu, 29 Jan 2015 15:08:39 -0800
Rian Hunter rian+suckless-...@thelig.ht wrote:
When MODE_INSERT is set we'd shift characters on the same
line forward before inserting our character in tputc().
This did not account for wide
Who here doesn't use ctags or similar? Because using navigation
techniques that only work if not technically enforced style guidelines
are followed is not really helpful to people who read and write code in
many different projects.
I don't use ctags. It's simple, if you use the correct code
The style(9)-changes were absolutely necessary and it's better to do this
as early as possible instead of waiting and waiting until it's too late
and you have a really big number of patches for a given program.
The thing I dislike most about the style changes is the alignment of
variable
Hi,
Hi Evan,
I'd prefer to stick to the existing makefile. It might be scary to
say that it depends on GNU make but bear in mind that of all the make
implementations I've tried only OpenBSD make had some trouble with it.
I think we should remove the GNU make dependence, because we don't
Applied!!!
system headers should come first, then a newline, then libc headers
then a newline then local headers.
I usually do just the inverse, first libc headers and later system
headers.
the libc headers are guaranteed to work regardless of the order of
inclusion but need to come after the system
types
=
user defined types start with a capital letter
when possible typedef struct {} Name;
Debatable. I'd like to hear more opinions on this.
For long time I followed the rule of kernel style of not using typedef
ever, but I changed it to the rule of using typedef only with
On Tue, Nov 18, 2014 at 09:53:33AM -0500, Greg Reagle wrote:
Another way to provide style guidance is to refer to a particular source
file or project as an epitome or paragon, if such a paragon exists That
takes much less time than explicitly developing a style guide.
Yes my initial idea
The Makefile made libutil.a and libutf.a identical. Here is a patch to fix
this.
-$(LIB): $(LIBUTFOBJ) $(LIBUTILOBJ)
+$(LIBUTF): $(LIBUTFOBJ)
+$(AR) -r -c $@ $?
+$(RANLIB) $@
+
+$(LIBUTIL): $(LIBUTILOBJ)
$(AR) -r -c $@ $?
$(RANLIB) $@
In fact, POSIX
types
=
user defined types start with a capital letter
when possible typedef struct {} Name;
Debatable. I'd like to hear more opinions on this.
In most suckless code types always start with capital letters (pretty
much acme/p9 influenced), as opposed to lower case everything else.
On 18 November 2014 19:25, Greg Reagle greg.rea...@umbc.edu wrote:
Hello. Is there a particular reason that a combination of cp, mkdir,
and chmod are used rather than the install command, in several suckless
projects (in the install target of Makefile)?
To me the answer is rather obvious.
In fact, POSIX says that it could be only:
$(LIB): libutf($(LIBUTFOBJ)) libutil($(LIBUTILOBJ))
but my experience is that it doesn't work in BSD systems :(
We now depend on GNU make. I am tired of fighting make implementations.
Ok, but I think current Makefile works in all the
Was there as reason or did the user trolls attending the last one
just have fun pissing off most of the suckless-contributors?
There's no reason to have a conference in Budapest when most people
come from a completely different part of the world.
However, I think we could put together some
C90, or any version of standard C, does not have a concept of system
headers, other than giving implementations permission to place their
own implementation-defined files in places searched by #include
h-char-sequence.
At this point I was talking about POSIX of course. C90 doesn't give
Good point. I'll apply it.
Thanks.
I have tested with your .zshrc and I don't have the problem you describe.
Are you sure that your config.h has the modifications the patch adds
to config.def.h?
Regards,
I did a quick survey of a number of other terminal emulators: tmux,
xterm, GNU screen, rxvt and dvtm all check the SHELL environment
variable. Checking SHELL is expected behaviour, and it's a useful one at
that.
Eric
Checking SHELL is common behaviour and any user setting SHELL will
Sorry, it is zsh.
It should work. I don't know why it doesn't work for you.
Do you have something in .zshrc?
Regards,
* Use strtonum() or estrtol() from sbase instead of atoi().
I agree here if the string comes from the user. If the string
is an internal string then there is no problem with atoi.
It is generally unlikely that the string has been validated to
be an integer before getting to atoi().
I am personally not opposed to using yacc. Yeah I like hand-rolled
recursive descent parsers are cool but yacc is a standard tool that
most people are familiar with.
Roberto, what do you think?
I agree with you. I think yacc is a suckless tool, and I am sure we will
have to use it for awk
So, it should be recognized as Delete key. What shell are you using?
Regards,
st - simple terminal
I asked you about your shell, not about your terminal emulator.
Regards,
* Use strtonum() or estrtol() from sbase instead of atoi().
I agree here if the string comes from the user. If the string
is an internal string then there is no problem with atoi.
Output of infocmp attached: out.txt
Output of ^VDelete:
^[[3~
You appear to have mistyped something. CSI requires only one bracket.
^[ is the ESC representation, so ^[[ is CSI. The terminal is generating
CSI 3 ~, that from your infocmp output is kdch1:
kdch1=\E[3~
So, it
If you are interested,
- Modify the keys generated by st. This option is not dificult, because
you
only have to modify the values in your config.h (I think I should put
this
option in the FAQ to).
If you are interested in this solution this is the patch:
diff --git
Hi,
I applied this patch and now when I press DELETE in the shell I get a
'~'.
Can you send the output of:
infocmp
and the output of
cat
^VDelete (press first control V and then Delete key)
Regards,
sent is interesting, but as a teacher, I need some mathematical symbols,
and a tool to write more than 3 words per slide...
(by the way, on debian, sent can't find any font...)
reveal.js is interesting too, but far from suckless...
I'll stick to markdown+pandoc-beamer for now until I
Hi folks,
my delete-key behaves as backspace since commit
9d9e049eac3dacb2725f9d792f7cdd2230062313 on all my linux machines
(debian, fedora, ubuntu) and openbsd.
i had this problem in the past (and wrote to this list about it), and
my .zshrc since then contains stty erase ^H:
It
Hi,
Above my opinion might be not constructive, so now I propose alternate
sequences:
I recommend a recall of Konsole style truecolor control seqences, and
propose to change them as follows:
CSI 38 ; 2 ; Pr ; Pg ; Pb m - CSI ? Pr ; Pg ; Pb $ m
CSI 48 ; 2 ; Pr ; Pg ; Pb m - CSI =
HOME must hold the value of the user's home directory, if this
value is not the value of /etc/passwd then you have an error
in your configuration. The same can be said about SHELL. You
can change your shell selection using chsh.
I don't agree with your interpretation of the POSIX verbiage,
Hi,
printf \x1b[38;2;255;100;0mTRUECOLOR\x1b[0m\n
I think this truecolor control sequence is not suckless.
Although I had already mentioned about this feature (konsole-style
true color support) on some places with my bad English, now I try to
say my opinion repeatedly in other
Ok as far as I can tell there are two unrelated issues involved (feel
free to correct me):
1) being able to use many colors (possibly all 2^24) at the same time
2) being able to set some of the used colors to RGB values
The first is not easily possible with ncurses. As you probably
Hi,
- POSIX states the SHELL environment variable ... shall represent a
pathname of the user's preferred command language interpreter. As
Well, this is a really complex topic. First because in this sentence
is not specified what is 'user's preferred command language
interpreter'. I read
in config.h something along the lines of
enum color_mode {
REVERSE,
COLOR,
};
struct selection_colors {
enum color_mode;
int colors[];
}
static const struct selection_colors same = { .color_mode = COLOR,
.colors = { 7, 0 } };
static const struct selection_colors different = {
Silvan Jegen said:
I do agree that this is the right approach. There is however another
instance of a wcwidth call on line st.c:3443 that should be handled as
well (maybe with abs in that particular case?).
As I get it, by the time wcwidth() is called there, all codepoints libc
is unaware
On 26 October 2014 19:36, Dmitrij D. Czarkoff czark...@gmail.com wrote:
Apparently no workaround is needed - just use sane libc or avoid using
new codepoints until glibc fixes that. Or, if you won't feel dirty
after that, send a patch to glibc instead.
'Just using musl' doesn't seem
Hi,
Apparently curses provides:
int init_color(short color, short r, short g, short b);
not quite sure what it does internally.
There is no support for true color in terminfo, so curses cannot
handle it. If your application need it the only way is to print
directly the sequences, and the
That's how POSIX-2001 defines it.
So yeah, using abs() to catch the invalid case is fine, but could be
refined even more.
Ok, it is true. We should catch this case, but I don't like the idea
of the abs. I think the correct solution should be print the invalid
character.
Regards,
A simpler possibility is to let it enabled by default and let the user
switch it off with a command line argument (like it's done for the mouse
grabbing). I think it would make sense as all major terminal emulators
support true color.
We have something similar for alternate screen to, but
Hi,
There is no support for true color in terminfo, so curses cannot
handle it.
Is this a fundamental limitation or just a lack of standardization?
https://lists.gnu.org/archive/html/bug-ncurses/2013-10/msg7.html
Mainly lack of standardization (this feature was added to
From: Roberto E. Vargas Caballero k...@shike2.com
---
Makefile |6 +-
config.def.h |2 ++
last.c | 59 ++
3 files changed, 66 insertions(+), 1 deletion(-)
create mode 100644 last.c
diff --git a/Makefile b/Makefile
From: Roberto E. Vargas Caballero k...@shike2.com
$? is expanded to all the requisites that are newer than the current target,
so we can avoid update files that were not modified since last time
the library was built.
---
Makefile |2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff
Patch letting st not embedding when given Window id is zero, without failing.
It makes sense to me, so I will apply it next week.
Regards,
On Sun, Oct 12, 2014, at 14:32, k...@shike2.com wrote:
That doesn't mean that the question of what the default should be is not
worth discussing.
Default configuration was discussed here some time ago, and suckless
developers agreed with current configuration. Both options, Backspace
Because that is the character sent by the key in this position with this
expected function (i.e. the x] key above enter) on DEC terminals. It's
First, terminals only send ascii codes, functions are assigned to them
by programs. Programs that are designed to work only with terminals
that
On Sat, Oct 11, 2014, at 04:07, k...@shike2.com wrote:
Value of erase key for example, or in general the configuration
of line kernel driver
These can't come from the profile either; since st opens a new tty that
is not the same device the user logged in on.
And the profile runs in the
Why shouldn't a non-login shell have a utmp session? And if this option
is to use a login shell, rather than merely using utmp, then I don't
think it should be a compile-time option - just because someone
sometimes wants a login shell (which could be done before, if desired,
by running e.g.
What exactly does terminal configuration mean here? TERM, LINES, and
COLUMNS? Shouldn't st itself be responsible for setting these? They
certainly don't belong in the profile.
Value of erase key for example, or in general the configuration
of line kernel driver (stty(1)). Backspace key in
-setenv(SHELL, sh, 1);
+setenv(SHELL, args[0], 1);
I'm not sure whether this is a good idea? The user asked to execute
args[0] instead of the shell, but this doesn't imply that args[0] is
his shell.
Yes, you are right, it should be setenv(SHELL, sh, 1);
Can you send a patch for
100 matches
Mail list logo
