[SCM] gawk branch, feature/minrx, updated. gawk-4.1.0-5923-g8cf5f867

[Arnold Robbins]

This is an automated email from the git hooks/post-receive script. It was
generated because a ref change was pushed to the repository containing
the project "gawk".

The branch, feature/minrx has been updated
       via  8cf5f86770ac91ed423c21abde5a1948d0925b04 (commit)
      from  832ce4bf9ee845aa3f3aa80d4e74100ecb62b14e (commit)

Those revisions listed above that are new to this repository have
not appeared on any other notification email; so we list those
revisions in full, below.

- Log -----------------------------------------------------------------
http://git.sv.gnu.org/cgit/gawk.git/commit/?id=8cf5f86770ac91ed423c21abde5a1948d0925b04

commit 8cf5f86770ac91ed423c21abde5a1948d0925b04
Author: Arnold D. Robbins <arnold@skeeve.com>
Date:   Tue Jan 28 14:50:27 2025 +0200

    Update minrx.cpp.

diff --git a/support/ChangeLog b/support/ChangeLog
index 9ee01ac6..f6a8df57 100644
--- a/support/ChangeLog
+++ b/support/ChangeLog
@@ -1,3 +1,7 @@
+2025-01-28         Arnold D. Robbins     <arnold@skeeve.com>
+
+       * minrx.cpp: Update again. More speedups.
+
 2025-01-27         Arnold D. Robbins     <arnold@skeeve.com>
 
        * minrx.cpp: Update again. More speedups.
diff --git a/support/minrx.cpp b/support/minrx.cpp
index dd9d0df0..319eb895 100644
--- a/support/minrx.cpp
+++ b/support/minrx.cpp
@@ -52,6 +52,12 @@
 #endif
 #include "minrx.h"
 
+#ifdef __GNUC__
+#define INLINE __attribute__((__always_inline__)) inline
+#else
+#define INLINE inline
+#endif
+
 #ifdef HAVE_CONFIG_H
 #include "config.h"
 #endif /* HAVE_CONFIG_H */
@@ -67,9 +73,9 @@
 
 namespace MinRX {
 
-template <typename UINT> inline auto ctz(UINT x) { return __builtin_ctz(x); }
-template <> inline auto ctz(unsigned long x) { return __builtin_ctzl(x); }
-template <> inline auto ctz(unsigned long long x) { return __builtin_ctzll(x); 
}
+template <typename UINT> auto ctz(UINT x) { return __builtin_ctz(x); }
+template <> auto ctz(unsigned long x) { return __builtin_ctzl(x); }
+template <> auto ctz(unsigned long long x) { return __builtin_ctzll(x); }
 
 template <typename TYPE, TYPE INIT = 0>
 struct COWVec {
@@ -172,15 +178,6 @@ struct COWVec {
                (*storage)[idx] = val;
                return *this;
        }
-       COWVec &sub(std::size_t idx, TYPE val) {
-               if (storage->refcnt > 1) {
-                       --storage->refcnt;
-                       storage = storage->clone();
-                       storage->refcnt = 1;
-               }
-               (*storage)[idx] -= val;
-               return *this;
-       }
 };
 
 template <typename UINT>
@@ -198,7 +195,7 @@ struct QSet {
                bits[0][0] = 0;
        }
        ~QSet() { ::operator delete(bits[0]); }
-       inline static std::uint64_t bit(UINT k) { return (std::uint64_t) 1 << 
(k & 0x3F); }
+       static std::uint64_t bit(UINT k) { return (std::uint64_t) 1 << (k & 
0x3F); }
        bool empty() const { return !bits[0][0]; }
        bool contains(UINT k) const {
                int i = 0, s = 6 * depth;
@@ -686,9 +683,6 @@ struct CSet {
        std::pair<std::optional<const std::array<bool, 256>>, 
std::optional<char>>
        firstbytes(WConv::Encoding e) const {
                std::array<bool, 256> fb = {};
-#ifdef ROARING
-               roaring_uint32_iterator_t *i;
-#endif
                auto firstunique = [](const std::array<bool, 256> &fb) -> 
std::optional<char> {
                        int n = 0, u = -1;
                        for (int i = 0; i < 256; ++i)
@@ -698,16 +692,6 @@ struct CSet {
                };
                switch (e) {
                case WConv::Encoding::Byte:
-#ifdef ROARING
-                       i = roaring_iterator_create(bitmap);
-                       while (i->has_value) {
-                               if (i->current_value > 255)
-                                       break;
-                               fb[i->current_value] = true;
-                               roaring_uint32_iterator_advance(i);
-                       }
-                       roaring_uint32_iterator_free(i);
-#else
                        for (const auto &r : ranges) {
                                if (r.min > 255)
                                        break;
@@ -715,23 +699,13 @@ struct CSet {
                                for (auto b = lo; b <= hi; b++)
                                        fb[b] = true;
                        }
-#endif
                        return {fb, firstunique(fb)};
                case WConv::Encoding::UTF8:
-#ifdef ROARING
-                       i = roaring_iterator_create(bitmap);
-                       while (i->has_value) {
-                               fb[utfprefix(i->current_value)] = true;
-                               roaring_uint32_iterator_advance(i);
-                       }
-                       roaring_uint32_iterator_free(i);
-#else
                        for (const auto &r : ranges) {
                                auto lo = utfprefix(r.min), hi = 
utfprefix(r.max);
                                for (auto b = lo; b <= hi; b++)
                                        fb[b] = true;
                        }
-#endif
                        return {fb, firstunique(fb)};
                default:
                        return {{}, {}};
@@ -1271,6 +1245,7 @@ struct Execute {
        const Node *nodes = r.nodes.data();
        Execute(const Regexp &r, minrx_regexec_flags_t flags, const char *bp, 
const char *ep) : r(r), flags(flags), wconv(r.enc, bp, ep) {}
        template <typename... XArgs>
+       INLINE
        void add(QVec<NInt, NState> &ncsv, NInt k, NInt nstk, const NState &ns, 
WChar wcnext, XArgs... xargs) {
                const Node &n = nodes[k];
                if (n.type <= Node::CSet) {
@@ -1475,7 +1450,7 @@ struct Execute {
                                auto [n, ns] = mcsvs[0].remove();
                                add(mcsvs[1], n + 1, nodes[n].nstk, ns, wcnext);
                        }
-                       if (!best.has_value() && (!r.firstcset.has_value() || 
r.firstcset->test(wcnext))) {
+                       if (!best.has_value() && (!r.firstbytes.has_value() || 
(wcnext != WConv::End && (*r.firstbytes)[(unsigned char) *wconv.cp]))) {
                                nsinit.boff = wconv.off();
                                add(mcsvs[1], 0, 0, nsinit, wcnext);
                        }
@@ -1495,7 +1470,7 @@ struct Execute {
                                auto [n, ns] = mcsvs[1].remove();
                                add(mcsvs[0], n + 1, nodes[n].nstk, ns, wcnext);
                        }
-                       if (!best.has_value() && (!r.firstcset.has_value() || 
r.firstcset->test(wcnext))) {
+                       if (!best.has_value() && (!r.firstbytes.has_value() || 
(wcnext != WConv::End && (*r.firstbytes)[(unsigned char) *wconv.cp]))) {
                                nsinit.boff = wconv.off();
                                add(mcsvs[0], 0, 0, nsinit, wcnext);
                        }

-----------------------------------------------------------------------

Summary of changes:
 support/ChangeLog |  4 ++++
 support/minrx.cpp | 51 +++++++++++++--------------------------------------
 2 files changed, 17 insertions(+), 38 deletions(-)


hooks/post-receive
-- 
gawk