Hi Bram,
2015-7-9(Thu) 18:30:55 UTC+9 Bram Moolenaar:
> Hirohito Higashi wrote:
>
> > Hi Bram and list!
> >
> > 2015-7-9(Thu) 10:47:12 UTC+9 h_east:
> > > Hi Bram!
> > >
> > > 2015-7-9(Thu) 4:55:17 UTC+9 Bram Moolenaar:
> > > > Hirohito Higashi wrote:
> > > >
> > > > > 2015-7-8(Wed) 2:08:03 UTC+9 Bram Moolenaar:
> > > > > > Hirohito Higashi wrote:
> > > > > >
> > > > > > > Hi Bram, Anatol and Vim developers!
> > > > > > >
> > > > > > > 2015-7-7(Tue) 8:20:43 UTC+9 Anatol Pomozov:
> > > > > > > > Is HEAD vim revision still broken? Could developers loos at this
> > > > > > > > bug? Many Arch developers report this issue and we would like to
> > > > > > > > release working version of vim.
> > > > > > >
> > > > > > > I think that is no need to correspond to the "rgba:".
> > > > > > > But, Vim is not good to output garbage being displayed which
> > > > > > > could not
> > > > > > > be interpreted the "Background color response".
> > > > > > >
> > > > > > > So, I wrote a patch.
> > > > > > > This patch is suppress to output garbage being displayed, when
> > > > > > > unknown
> > > > > > > background color response received. (e.g.
> > > > > > > "11;rgba:aaaa/rrrr/gggg/bbbb")
> > > > > > >
> > > > > > > Please confirm this.
> > > > > >
> > > > > > It's a bit tricky.
> > > > > >
> > > > > > > diff -r bfc3682510d6 src/term.c
> > > > > > > --- a/src/term.c Sat Jul 04 15:05:14 2015 +0200
> > > > > > > +++ b/src/term.c Tue Jul 07 13:08:17 2015 +0900
> > > > > > > @@ -4272,7 +4272,7 @@
> > > > > > >
> > > > > > > if ((*T_CRV != NUL || *T_U7 != NUL || *T_RBG != NUL)
> > > > > > > && ((tp[0] == ESC && tp[1] == '[' && len >= 3)
> > > > > > > - || (tp[0] == ESC && tp[1] == ']' && len >=
> > > > > > > 24)
> > > > > > > + || (tp[0] == ESC && tp[1] == ']' && len >=
> > > > > > > 3)
> > > > > > > || (tp[0] == CSI && len >= 2))
> > > > > > > && (VIM_ISDIGIT(*argp) || *argp == '>' || *argp
> > > > > > > == '?'))
> > > > > > > {
> > > > > >
> > > > > > I think the length check should be len >= 5, since that's what is
> > > > > > used
> > > > > > below.
> > > > >
> > > > > No problem. Because it checked in src/term.c:L4414.
> > > > > (Note: Line number is the after applying my patch)
> > > >
> > > > But then we will search for the whole escape sequence.
> > > >
> > > > > 4413 else if (*T_RBG != NUL
> > > > > 4414 && len >= 5 - (tp[0] == CSI)
> > > > > 4415 && argp[0] == '1' && argp[1] == '1' &&
> > > > > argp[2] == ';')
> > > > > 4416 {
> > > > >
> > > > > >
> > > > > >
> > > > > > > + j = 2 - (tp[0] == CSI);
> > > > > > > + for (i = j; i < len; ++i)
> > > > > > > + if (tp[i] == '\007'
> > > > > > > + || (tp[0] == CSI ? tp[i] == STERM
> > > > > > > + : i + 1 < len && tp[i] == ESC && tp[i +
> > > > > > > 1] == '\\'))
> > > > > > > + {
> > > > > > > + LOG_TR("Received unknown RBG");
> > > > > > > + key_name[0] = (int)KS_EXTRA;
> > > > > > > + key_name[1] = (int)KE_IGNORE;
> > > > > > > + slen = i + 1 + (tp[i] == ESC);
> > > > > > > + break;
> > > > > > > + }
> > > > > > > +
> > > > > > > + if (i == len)
> > > > > > > + {
> > > > > > > + LOG_TR("not enough characters for RB");
> > > > > > > + return -1; /* not enough
> > > > > > > characters */
> > > > > > > + }
> > > > > > > + }
> > > > > >
> > > > > > If the terminating character is missing, or the specific terminal
> > > > > > uses
> > > > > > another character to terminate the sequence, then this will eat
> > > > > > everything.
> > > > >
> > > > > This patch discard only the following pattern. ('*' is any character
> > > > > string)
> > > > > ESC]11;*BEL
> > > > > ESC]11;*ESC\
> > > > > CSI11;*BEL
> > > > > CSI11;*ST
> > > > >
> > > > > So, I think the problem is less likely to occur.
> > > >
> > > > But when we get ESC]11; and then no BEL or ESC\ follows, we never get
> > > > out of this loop, we run into the "not enough characters" part forever.
> > >
> > > Sorry. I have big mistake.
> > > Patch 7.4.757 has confused the CSI(0x9b) and OSC(0x9d).
> > > And, I did not notice it.
> > >
> > > Please wait for a few days, because I refactor the term.c.
> >
> > Done.
> >
> > Changed contents:
> > - Defined the OSC in the same way as CSI, DCS.
> > - CSI, OSC and DCS check process was clearly separated.
> >
> > Please confirm again.
>
> Thanks, I'll look into it.
>
> > BTW...
> > > > But when we get ESC]11; and then no BEL or ESC\ follows, we never get
> > > > out of this loop, we run into the "not enough characters" part forever.
> >
> > This problem is also in the process of the current of CSI and DCS.
> > Find and check for below string in term.c
> > "Not enough characters for CRV"
> > "not enough characters for XT"
>
> We might need to find a generic solution. A timeout is probably
> appropriate, these escape sequences should be completely received within
> a short time. Although on a slow connection it might get split (the
> "entering a tunnel" problem).
Sure.
This matter is corresponding later.
I updated a patch. (BELL termination check were missing for RBG)
And I received a review from Hayaki Saito.
He says:
- 8-bit response can not received. (Contains mouse input)
- may_req_bg_color() call timing is too early. Should Called after t_CRV
response received and switch to 8-bit mode.
The findings, I would try to correspond to the weekend.
These are the issues from before.
So, it might be good to include rbg_fix5.patch first :-)
Thanks.
--
Best regards,
Hirohito Higashi (a.k.a h_east)
--
--
You received this message from the "vim_dev" maillist.
Do not top-post! Type your reply below the text you are replying to.
For more information, visit http://www.vim.org/maillist.php
---
You received this message because you are subscribed to the Google Groups
"vim_dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
diff --git a/src/ascii.h b/src/ascii.h
--- a/src/ascii.h
+++ b/src/ascii.h
@@ -37,6 +37,7 @@
#define CSI 0x9b /* Control Sequence Introducer */
#define CSI_STR "\233"
#define DCS 0x90 /* Device Control String */
+#define OSC 0x9d /* Operating System Command */
#define STERM 0x9c /* String Terminator */
#define POUND 0xA3
@@ -121,6 +122,7 @@
#define CSI 0x9b /* Control Sequence Introducer */
#define CSI_STR "\233"
#define DCS 0x90 /* Device Control String */
+#define OSC 0x9d /* Operating System Command */
#define STERM 0x9c /* String Terminator */
#define POUND 0xB1
diff --git a/src/term.c b/src/term.c
--- a/src/term.c
+++ b/src/term.c
@@ -2364,7 +2364,7 @@
if (p[1] == '[')
return CSI;
if (p[1] == ']')
- return 0x9d;
+ return OSC;
if (p[1] == 'O')
return 0x8f;
}
@@ -4261,18 +4261,11 @@
* - Cursor position report: <Esc>[{row};{col}R
* The final byte must be 'R'. It is used for checking the
* ambiguous-width character state.
- *
- * - Background color response:
- * <Esc>]11;rgb:{rrrr}/{gggg}/{bbbb}\007
- * Or
- * <Esc>]11;rgb:{rrrr}/{gggg}/{bbbb}ST
- * The final byte must be '\007' or ST(0x9c or ESC\).
*/
char_u *argp = tp[0] == CSI ? tp + 1 : tp + 2;
- if ((*T_CRV != NUL || *T_U7 != NUL || *T_RBG != NUL)
+ if ((*T_CRV != NUL || *T_U7 != NUL)
&& ((tp[0] == ESC && tp[1] == '[' && len >= 3)
- || (tp[0] == ESC && tp[1] == ']' && len >= 24)
|| (tp[0] == CSI && len >= 2))
&& (VIM_ISDIGIT(*argp) || *argp == '>' || *argp == '?'))
{
@@ -4410,34 +4403,57 @@
key_name[1] = (int)KE_IGNORE;
slen = i + 1;
}
- else if (*T_RBG != NUL
- && len >= 24 - (tp[0] == CSI)
- && len >= 24 - (tp[0] == CSI) + (argp[21] == ESC)
- && argp[0] == '1' && argp[1] == '1'
- && argp[2] == ';' && argp[3] == 'r' && argp[4] == 'g'
- && argp[5] == 'b' && argp[6] == ':'
- && argp[11] == '/' && argp[16] == '/'
- && (argp[21] == '\007' || argp[21] == STERM
- || (argp[21] == ESC && argp[22] == '\\')))
+ }
+
+ /* Check for some responses from the terminal starting with
+ * "<Esc>]" or OSC:
+ *
+ * - Background color response:
+ * <Esc>]11;rgb:{rrrr}/{gggg}/{bbbb}\007
+ * Or
+ * <Esc>]11;rgb:{rrrr}/{gggg}/{bbbb}ST
+ * The final byte must be '\007' or ST(0x9c or ESC\).
+ */
+ else if (*T_RBG != NUL
+ && ((tp[0] == ESC && tp[1] == ']' && len >= 2)
+ || tp[0] == OSC))
+ {
+ j = 1 + (tp[0] != OSC);
+ for (i = j; i < len; ++i)
+ if (tp[i] == '\007' || (tp[0] == OSC ? tp[i] == STERM
+ : (tp[i] == ESC && tp[i + 1] == '\\' && i + 1 < len)))
+ {
+ if (i - j >= 21 && STRNCMP(tp + j, "11;rgb:", 7) == 0
+ && tp[j + 11] == '/' && tp[j + 16] == '/'
+ && !option_was_set((char_u *)"bg"))
+ {
+ LOG_TR("Received RBG");
+ rbg_status = RBG_GOT;
+ set_option_value((char_u *)"bg", 0L, (char_u *)(
+ (3 * '6' < tp[j+7] + tp[j+12] + tp[j+17])
+ ? "light" : "dark"), 0);
+ reset_option_was_set((char_u *)"bg");
+ redraw_asap(CLEAR);
+ }
+
+ key_name[0] = (int)KS_EXTRA;
+ key_name[1] = (int)KE_IGNORE;
+ slen = i + 1 + (tp[i] == ESC);
+ break;
+ }
+ if (i == len)
{
- LOG_TR("Received RBG");
- rbg_status = RBG_GOT;
- if (!option_was_set((char_u *)"bg"))
- {
- set_option_value((char_u *)"bg", 0L, (char_u *)(
- (3 * '6' < argp[7] + argp[12] + argp[17])
- ? "light" : "dark"), 0);
- reset_option_was_set((char_u *)"bg");
- redraw_asap(CLEAR);
- }
- key_name[0] = (int)KS_EXTRA;
- key_name[1] = (int)KE_IGNORE;
- slen = 24 - (tp[0] == CSI) + (argp[21] == ESC);
+ LOG_TR("not enough characters for RB");
+ return -1; /* not enough characters */
}
}
- /* Check for '<Esc>P1+r<hex bytes><Esc>\'. A "0" instead of the
- * "1" means an invalid request. */
+ /* Check for some responses from the terminal starting with
+ * "<Esc>P" or DCS:
+ *
+ * - key code response from xterm: <Esc>P1+r<name>=<string><Esc>\
+ * "0" instead of the "1" means an invalid request.
+ */
else if (check_for_codes
&& ((tp[0] == ESC && tp[1] == 'P' && len >= 2)
|| tp[0] == DCS))
