bug#40407: [PATCH] slow ENCODE_FILE and DECODE_FILE

["Mattias Engdegård" <mattiase@acm.org>]

[-- Attachment #1: Type: text/plain, Size: 1158 bytes --]

4 apr. 2020 kl. 11.26 skrev Eli Zaretskii <eliz@gnu.org>:

> DECODE_FILE is called because the file name in question starts with a
> "~"?  Otherwise, I don't think I understand why would expand-file-name
> need to decode a file name.

Maybe it's because default-directory started with a tilde. It doesn't really matter; it's a common case, and the profiler tells us as much.

> IME, the cases where we can safely assume it's OK to return the same
> string are actually very rare.  It is no accident that you saw so few
> calls of these functions where we use that optional behavior.

This does not mean that the remaining 179 calls require a copy; they just use the default value of the parameter.

> Neither, IMO.  Again, it's a separate problem, and let's keep our
> sights squarely on the original issue you wanted to fix.  Let's tackle
> the NOCOPY issue in a separate discussion, OK?

Thank you, a separate bug for it is fine.

Here is a revised patch which takes the nocopy parameter into account (in its inverted sense). Obviously it needs to be adapted if the nocopy inversion is dealt with first; the two bugs do not commute.


[-- Attachment #2: 0001-Avoid-expensive-recoding-for-ASCII-identity-cases-bu.patch --]
[-- Type: application/octet-stream, Size: 2146 bytes --]

From 0c6139ab490733f3c1257665535fc4ed2ad0dbe7 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Mattias=20Engdeg=C3=A5rd?= <mattiase@acm.org>
Date: Fri, 3 Apr 2020 16:01:01 +0200
Subject: [PATCH] Avoid expensive recoding for ASCII identity cases (bug#40407)

Optimise for the common case of encoding or decoding an ASCII-only
string using an ASCII-compatible coding, for file names in particular.

* src/coding.c (string_ascii_p): New function.
(code_convert_string): Return the input string for ASCII-only inputs
and ASCII-compatible codings.
---
 src/coding.c | 23 ++++++++++++++++++++++-
 1 file changed, 22 insertions(+), 1 deletion(-)

diff --git a/src/coding.c b/src/coding.c
index 0bea2a0c2b..0fdbc95939 100644
--- a/src/coding.c
+++ b/src/coding.c
@@ -9471,6 +9471,17 @@ used (which may be different from CODING-SYSTEM if CODING-SYSTEM is
   return code_convert_region (start, end, coding_system, destination, 1, 0);
 }
 
+/* Whether a (unibyte) string only contains chars in the 0..127 range.  */
+static bool
+string_ascii_p (Lisp_Object str)
+{
+  ptrdiff_t nbytes = SBYTES (str);
+  for (ptrdiff_t i = 0; i < nbytes; i++)
+    if (SREF (str, i) > 127)
+      return false;
+  return true;
+}
+
 Lisp_Object
 code_convert_string (Lisp_Object string, Lisp_Object coding_system,
 		     Lisp_Object dst_object, bool encodep, bool nocopy,
@@ -9502,7 +9513,17 @@ code_convert_string (Lisp_Object string, Lisp_Object coding_system,
   chars = SCHARS (string);
   bytes = SBYTES (string);
 
-  if (BUFFERP (dst_object))
+  if (EQ (dst_object, Qt))
+    {
+      /* Fast path for ASCII-only input and an ASCII-compatible coding:
+         act as identity.  */
+      Lisp_Object attrs = CODING_ID_ATTRS (coding.id);
+      if (! NILP (CODING_ATTR_ASCII_COMPAT (attrs))
+          && (STRING_MULTIBYTE (string)
+              ? (chars == bytes) : string_ascii_p (string)))
+	return nocopy ? Fcopy_sequence (string) : string;
+    }
+  else if (BUFFERP (dst_object))
     {
       struct buffer *buf = XBUFFER (dst_object);
       ptrdiff_t buf_pt = BUF_PT (buf);
-- 
2.21.1 (Apple Git-122.3)