Misc cleanups:

- built-in regex is required only when ${USE_WIDECHAR} == "yes"

- -DUSE_WIDECHAR cpp flag should be determined by ${USE_WIDECHAR},
not ${USE_BUILTIN_REGEX}

- reduce .if -- .endif block for clarity

Put back local regex (thanks Rin)

we don't need the extra copy wide-regex anymore.

Don't let clang warn about comparing pointers with string literals.

Remove no longer needed workaround for PR/54656.

PR/54656: Martin Husemann: gcc-8 miscompiles ex_filter.c even with -O1, make
it -O0. The tm argument gets passed in as non-null and it is non-null after
the command execution, but then becomes null and causes a null dereference.

Add missing break.

- use -Wno-error=implicit-fallthrough with GCC7.

Add imctrl and imkey options, inspired by cannactrl and fepkey options in
nvi-m17n by itojun.

If imctrl option is set, input method is controlled by using escape
sequences compatible to Tera Term and RLogin. The state of input method in
commands specified by imkey option is saved and restored automatically.
Input method is then deactivated on returning to command mode.

branches: 1.16.2; 1.16.4;
Bump WARNS to 5. Add -Wno-old-style-definition to regex.c for gcc.

Enable -Wuninitialized -Wformat-security for clang.

We always need -Wno-format-nonliteral for v_increment.c whether compiler is
gcc or clang.

Cosmetic changes. No functional changes.

Sorry, remove illegal indents. Real fix for build with USE_WIDECHAR != "yes".

- Add -Wno-unused and -Wno-unsequenced for gcc and clang, respectively.
This is because encoding conversion macros yield such like:

#define FILE2INT5(sp,buf,n,nlen,w,wlen) (w = n, wlen = nlen, 0)

- Silence gcc warnings on nonliteral arguments for snprintf(3) in v_increment.c,
that are overlooked for swprintf(3) in the case of USE_WIDECHAR == "yes".

Use mvwchgat(3), which is not a macro in general (and in our case).

No need to compile in ip_*.c and ipc_*.c to nvi, although they are
necessary for generation of header files.

Reduce 5% of binary size on amd64.

Introduce HAVE_XXX macros to where they should be.
No binary changes, but possibly improve portability for, e.g.,
sync with pkgsrc version in the near future.

PR/50484: Rin Okuyama: fix the script command of vi(1)

add QUEUEDEBUG commented out

branches: 1.5.2; 1.5.4; 1.5.8;
PR toolchain/48833: Ilia Zykov: nvi build fails if under /common

When autogenerating headers from source (yuck) use a more selective
pattern to avoid selecting anything if part of the current absolute
path happens to match part of the nvi source tree.

branches: 1.4.2; 1.4.4;
Don't rebuild the header files all the time if they did not change

Bring back the compatibility db3->db1 code from the old vi.
This is selected with -DUSE_DB1 in the Makefile. By default
use use the db1 specific code from the current nvi which is
enabled using -DUSE_BUNDLED_DB. Unfortunately recovery seems
to be broken for both.

fix tracing.

Add build glue and docs

Don't let clang warn about comparing pointers with string literals.

Remove no longer needed workaround for PR/54656.

PR/54656: Martin Husemann: gcc-8 miscompiles ex_filter.c even with -O1, make
it -O0. The tm argument gets passed in as non-null and it is non-null after
the command execution, but then becomes null and causes a null dereference.

Add missing break.

- use -Wno-error=implicit-fallthrough with GCC7.

Add imctrl and imkey options, inspired by cannactrl and fepkey options in
nvi-m17n by itojun.

If imctrl option is set, input method is controlled by using escape
sequences compatible to Tera Term and RLogin. The state of input method in
commands specified by imkey option is saved and restored automatically.
Input method is then deactivated on returning to command mode.

branches: 1.16.2; 1.16.4;
Bump WARNS to 5. Add -Wno-old-style-definition to regex.c for gcc.

Enable -Wuninitialized -Wformat-security for clang.

We always need -Wno-format-nonliteral for v_increment.c whether compiler is
gcc or clang.

Cosmetic changes. No functional changes.

Sorry, remove illegal indents. Real fix for build with USE_WIDECHAR != "yes".

- Add -Wno-unused and -Wno-unsequenced for gcc and clang, respectively.
This is because encoding conversion macros yield such like:

#define FILE2INT5(sp,buf,n,nlen,w,wlen) (w = n, wlen = nlen, 0)

- Silence gcc warnings on nonliteral arguments for snprintf(3) in v_increment.c,
that are overlooked for swprintf(3) in the case of USE_WIDECHAR == "yes".

Use mvwchgat(3), which is not a macro in general (and in our case).

No need to compile in ip_*.c and ipc_*.c to nvi, although they are
necessary for generation of header files.

Reduce 5% of binary size on amd64.

Introduce HAVE_XXX macros to where they should be.
No binary changes, but possibly improve portability for, e.g.,
sync with pkgsrc version in the near future.

PR/50484: Rin Okuyama: fix the script command of vi(1)

add QUEUEDEBUG commented out

branches: 1.5.2; 1.5.4; 1.5.8;
PR toolchain/48833: Ilia Zykov: nvi build fails if under /common

When autogenerating headers from source (yuck) use a more selective
pattern to avoid selecting anything if part of the current absolute
path happens to match part of the nvi source tree.

branches: 1.4.2; 1.4.4;
Don't rebuild the header files all the time if they did not change

Bring back the compatibility db3->db1 code from the old vi.
This is selected with -DUSE_DB1 in the Makefile. By default
use use the db1 specific code from the current nvi which is
enabled using -DUSE_BUNDLED_DB. Unfortunately recovery seems
to be broken for both.

fix tracing.

Add build glue and docs

Remove no longer needed workaround for PR/54656.

PR/54656: Martin Husemann: gcc-8 miscompiles ex_filter.c even with -O1, make
it -O0. The tm argument gets passed in as non-null and it is non-null after
the command execution, but then becomes null and causes a null dereference.

Add missing break.

- use -Wno-error=implicit-fallthrough with GCC7.

Add imctrl and imkey options, inspired by cannactrl and fepkey options in
nvi-m17n by itojun.

If imctrl option is set, input method is controlled by using escape
sequences compatible to Tera Term and RLogin. The state of input method in
commands specified by imkey option is saved and restored automatically.
Input method is then deactivated on returning to command mode.

branches: 1.16.2; 1.16.4;
Bump WARNS to 5. Add -Wno-old-style-definition to regex.c for gcc.

Enable -Wuninitialized -Wformat-security for clang.

We always need -Wno-format-nonliteral for v_increment.c whether compiler is
gcc or clang.

Cosmetic changes. No functional changes.

Sorry, remove illegal indents. Real fix for build with USE_WIDECHAR != "yes".

- Add -Wno-unused and -Wno-unsequenced for gcc and clang, respectively.
This is because encoding conversion macros yield such like:

#define FILE2INT5(sp,buf,n,nlen,w,wlen) (w = n, wlen = nlen, 0)

- Silence gcc warnings on nonliteral arguments for snprintf(3) in v_increment.c,
that are overlooked for swprintf(3) in the case of USE_WIDECHAR == "yes".

Use mvwchgat(3), which is not a macro in general (and in our case).

No need to compile in ip_*.c and ipc_*.c to nvi, although they are
necessary for generation of header files.

Reduce 5% of binary size on amd64.

Introduce HAVE_XXX macros to where they should be.
No binary changes, but possibly improve portability for, e.g.,
sync with pkgsrc version in the near future.

PR/50484: Rin Okuyama: fix the script command of vi(1)

add QUEUEDEBUG commented out

branches: 1.5.2; 1.5.4; 1.5.8;
PR toolchain/48833: Ilia Zykov: nvi build fails if under /common

When autogenerating headers from source (yuck) use a more selective
pattern to avoid selecting anything if part of the current absolute
path happens to match part of the nvi source tree.

branches: 1.4.2; 1.4.4;
Don't rebuild the header files all the time if they did not change

Bring back the compatibility db3->db1 code from the old vi.
This is selected with -DUSE_DB1 in the Makefile. By default
use use the db1 specific code from the current nvi which is
enabled using -DUSE_BUNDLED_DB. Unfortunately recovery seems
to be broken for both.

fix tracing.

Add build glue and docs

PR/54656: Martin Husemann: gcc-8 miscompiles ex_filter.c even with -O1, make
it -O0. The tm argument gets passed in as non-null and it is non-null after
the command execution, but then becomes null and causes a null dereference.

Add missing break.

- use -Wno-error=implicit-fallthrough with GCC7.

Add imctrl and imkey options, inspired by cannactrl and fepkey options in
nvi-m17n by itojun.

If imctrl option is set, input method is controlled by using escape
sequences compatible to Tera Term and RLogin. The state of input method in
commands specified by imkey option is saved and restored automatically.
Input method is then deactivated on returning to command mode.

branches: 1.16.2; 1.16.4;
Bump WARNS to 5. Add -Wno-old-style-definition to regex.c for gcc.

Enable -Wuninitialized -Wformat-security for clang.

We always need -Wno-format-nonliteral for v_increment.c whether compiler is
gcc or clang.

Cosmetic changes. No functional changes.

Sorry, remove illegal indents. Real fix for build with USE_WIDECHAR != "yes".

- Add -Wno-unused and -Wno-unsequenced for gcc and clang, respectively.
This is because encoding conversion macros yield such like:

#define FILE2INT5(sp,buf,n,nlen,w,wlen) (w = n, wlen = nlen, 0)

- Silence gcc warnings on nonliteral arguments for snprintf(3) in v_increment.c,
that are overlooked for swprintf(3) in the case of USE_WIDECHAR == "yes".

Use mvwchgat(3), which is not a macro in general (and in our case).

No need to compile in ip_*.c and ipc_*.c to nvi, although they are
necessary for generation of header files.

Reduce 5% of binary size on amd64.

Introduce HAVE_XXX macros to where they should be.
No binary changes, but possibly improve portability for, e.g.,
sync with pkgsrc version in the near future.

PR/50484: Rin Okuyama: fix the script command of vi(1)

add QUEUEDEBUG commented out

branches: 1.5.2; 1.5.4; 1.5.8;
PR toolchain/48833: Ilia Zykov: nvi build fails if under /common

When autogenerating headers from source (yuck) use a more selective
pattern to avoid selecting anything if part of the current absolute
path happens to match part of the nvi source tree.

branches: 1.4.2; 1.4.4;
Don't rebuild the header files all the time if they did not change

Bring back the compatibility db3->db1 code from the old vi.
This is selected with -DUSE_DB1 in the Makefile. By default
use use the db1 specific code from the current nvi which is
enabled using -DUSE_BUNDLED_DB. Unfortunately recovery seems
to be broken for both.

fix tracing.

Add build glue and docs

Add missing break.

- use -Wno-error=implicit-fallthrough with GCC7.

Add imctrl and imkey options, inspired by cannactrl and fepkey options in
nvi-m17n by itojun.

If imctrl option is set, input method is controlled by using escape
sequences compatible to Tera Term and RLogin. The state of input method in
commands specified by imkey option is saved and restored automatically.
Input method is then deactivated on returning to command mode.

branches: 1.16.2; 1.16.4;
Bump WARNS to 5. Add -Wno-old-style-definition to regex.c for gcc.

Enable -Wuninitialized -Wformat-security for clang.

We always need -Wno-format-nonliteral for v_increment.c whether compiler is
gcc or clang.

Cosmetic changes. No functional changes.

Sorry, remove illegal indents. Real fix for build with USE_WIDECHAR != "yes".

- Add -Wno-unused and -Wno-unsequenced for gcc and clang, respectively.
This is because encoding conversion macros yield such like:

#define FILE2INT5(sp,buf,n,nlen,w,wlen) (w = n, wlen = nlen, 0)

- Silence gcc warnings on nonliteral arguments for snprintf(3) in v_increment.c,
that are overlooked for swprintf(3) in the case of USE_WIDECHAR == "yes".

Use mvwchgat(3), which is not a macro in general (and in our case).

No need to compile in ip_*.c and ipc_*.c to nvi, although they are
necessary for generation of header files.

Reduce 5% of binary size on amd64.

Introduce HAVE_XXX macros to where they should be.
No binary changes, but possibly improve portability for, e.g.,
sync with pkgsrc version in the near future.

PR/50484: Rin Okuyama: fix the script command of vi(1)

add QUEUEDEBUG commented out

branches: 1.5.2; 1.5.4; 1.5.8;
PR toolchain/48833: Ilia Zykov: nvi build fails if under /common

When autogenerating headers from source (yuck) use a more selective
pattern to avoid selecting anything if part of the current absolute
path happens to match part of the nvi source tree.

branches: 1.4.2; 1.4.4;
Don't rebuild the header files all the time if they did not change

Bring back the compatibility db3->db1 code from the old vi.
This is selected with -DUSE_DB1 in the Makefile. By default
use use the db1 specific code from the current nvi which is
enabled using -DUSE_BUNDLED_DB. Unfortunately recovery seems
to be broken for both.

fix tracing.

Add build glue and docs

- use -Wno-error=implicit-fallthrough with GCC7.

Add imctrl and imkey options, inspired by cannactrl and fepkey options in
nvi-m17n by itojun.

If imctrl option is set, input method is controlled by using escape
sequences compatible to Tera Term and RLogin. The state of input method in
commands specified by imkey option is saved and restored automatically.
Input method is then deactivated on returning to command mode.

branches: 1.16.2;
Bump WARNS to 5. Add -Wno-old-style-definition to regex.c for gcc.

Enable -Wuninitialized -Wformat-security for clang.

We always need -Wno-format-nonliteral for v_increment.c whether compiler is
gcc or clang.

Cosmetic changes. No functional changes.

Sorry, remove illegal indents. Real fix for build with USE_WIDECHAR != "yes".

- Add -Wno-unused and -Wno-unsequenced for gcc and clang, respectively.
This is because encoding conversion macros yield such like:

#define FILE2INT5(sp,buf,n,nlen,w,wlen) (w = n, wlen = nlen, 0)

- Silence gcc warnings on nonliteral arguments for snprintf(3) in v_increment.c,
that are overlooked for swprintf(3) in the case of USE_WIDECHAR == "yes".

Use mvwchgat(3), which is not a macro in general (and in our case).

No need to compile in ip_*.c and ipc_*.c to nvi, although they are
necessary for generation of header files.

Reduce 5% of binary size on amd64.

Introduce HAVE_XXX macros to where they should be.
No binary changes, but possibly improve portability for, e.g.,
sync with pkgsrc version in the near future.

PR/50484: Rin Okuyama: fix the script command of vi(1)

add QUEUEDEBUG commented out

branches: 1.5.2; 1.5.4; 1.5.8;
PR toolchain/48833: Ilia Zykov: nvi build fails if under /common

When autogenerating headers from source (yuck) use a more selective
pattern to avoid selecting anything if part of the current absolute
path happens to match part of the nvi source tree.

branches: 1.4.2; 1.4.4;
Don't rebuild the header files all the time if they did not change

Bring back the compatibility db3->db1 code from the old vi.
This is selected with -DUSE_DB1 in the Makefile. By default
use use the db1 specific code from the current nvi which is
enabled using -DUSE_BUNDLED_DB. Unfortunately recovery seems
to be broken for both.

fix tracing.

Add build glue and docs

Bump WARNS to 5. Add -Wno-old-style-definition to regex.c for gcc.

Enable -Wuninitialized -Wformat-security for clang.

We always need -Wno-format-nonliteral for v_increment.c whether compiler is
gcc or clang.

Cosmetic changes. No functional changes.

Sorry, remove illegal indents. Real fix for build with USE_WIDECHAR != "yes".

- Add -Wno-unused and -Wno-unsequenced for gcc and clang, respectively.
This is because encoding conversion macros yield such like:

#define FILE2INT5(sp,buf,n,nlen,w,wlen) (w = n, wlen = nlen, 0)

- Silence gcc warnings on nonliteral arguments for snprintf(3) in v_increment.c,
that are overlooked for swprintf(3) in the case of USE_WIDECHAR == "yes".

Use mvwchgat(3), which is not a macro in general (and in our case).

No need to compile in ip_*.c and ipc_*.c to nvi, although they are
necessary for generation of header files.

Reduce 5% of binary size on amd64.

Introduce HAVE_XXX macros to where they should be.
No binary changes, but possibly improve portability for, e.g.,
sync with pkgsrc version in the near future.

PR/50484: Rin Okuyama: fix the script command of vi(1)

add QUEUEDEBUG commented out

branches: 1.5.2; 1.5.4; 1.5.8;
PR toolchain/48833: Ilia Zykov: nvi build fails if under /common

When autogenerating headers from source (yuck) use a more selective
pattern to avoid selecting anything if part of the current absolute
path happens to match part of the nvi source tree.

branches: 1.4.2; 1.4.4;
Don't rebuild the header files all the time if they did not change

Bring back the compatibility db3->db1 code from the old vi.
This is selected with -DUSE_DB1 in the Makefile. By default
use use the db1 specific code from the current nvi which is
enabled using -DUSE_BUNDLED_DB. Unfortunately recovery seems
to be broken for both.

fix tracing.

Add build glue and docs

No need to compile in ip_*.c and ipc_*.c to nvi, although they are
necessary for generation of header files.

Reduce 5% of binary size on amd64.

Introduce HAVE_XXX macros to where they should be.
No binary changes, but possibly improve portability for, e.g.,
sync with pkgsrc version in the near future.

PR/50484: Rin Okuyama: fix the script command of vi(1)

add QUEUEDEBUG commented out

branches: 1.5.2; 1.5.4; 1.5.8;
PR toolchain/48833: Ilia Zykov: nvi build fails if under /common

When autogenerating headers from source (yuck) use a more selective
pattern to avoid selecting anything if part of the current absolute
path happens to match part of the nvi source tree.

branches: 1.4.2; 1.4.4;
Don't rebuild the header files all the time if they did not change

Bring back the compatibility db3->db1 code from the old vi.
This is selected with -DUSE_DB1 in the Makefile. By default
use use the db1 specific code from the current nvi which is
enabled using -DUSE_BUNDLED_DB. Unfortunately recovery seems
to be broken for both.

fix tracing.

Add build glue and docs

PR/50484: Rin Okuyama: fix the script command of vi(1)

add QUEUEDEBUG commented out

branches: 1.5.2; 1.5.4;
PR toolchain/48833: Ilia Zykov: nvi build fails if under /common

When autogenerating headers from source (yuck) use a more selective
pattern to avoid selecting anything if part of the current absolute
path happens to match part of the nvi source tree.

branches: 1.4.2; 1.4.4;
Don't rebuild the header files all the time if they did not change

Bring back the compatibility db3->db1 code from the old vi.
This is selected with -DUSE_DB1 in the Makefile. By default
use use the db1 specific code from the current nvi which is
enabled using -DUSE_BUNDLED_DB. Unfortunately recovery seems
to be broken for both.

fix tracing.

Add build glue and docs