- Dillo-dev - lists.mailman3.com

Re: Issue 187: "Configure should stop when no compiler is found"

by a1ex＠dismail.de

Rodrigo Arias <rodarima(a)gmail.com> wrote: > I think something like this may be more strict, as we test that it > can compile a simple C++ program: > > diff --git a/configure.ac b/configure.ac > index 5c308cb7..7d156b81 100644 > --- a/configure.ac > +++ b/configure.ac > @@ -129,6 +129,19 @@ AC_PROG_CXX > AC_PROG_RANLIB > AC_PROG_CPP > > +AC_MSG_CHECKING(if C++ compiler '$CXX' works) > +AC_LANG_PUSH([C++]) > +AC_COMPILE_IFELSE([AC_LANG_PROGRAM([[ > + #ifndef __cplusplus > + #error "No C++ support, AC_PROG_CXX failed" > + #endif > + ]])], > + [AC_MSG_RESULT([yes])], > + [AC_MSG_RESULT(no) > + AC_MSG_FAILURE([C++ compiler doesn't work])] > +) > +AC_LANG_POP([C++]) > + > dnl ---------------------------- > dnl Check our char and int types > dnl ---------------------------- Very nice. Works fine here :)

Oct. 5, 2024

Re: CSS extra units

by Rodrigo Arias

On Sat, Oct 05, 2024 at 06:25:20PM +0200, a1ex(a)dismail.de wrote: >After a/b testing this with a fair amount of sites, I think it's a good >PR. Maybe 10-20% of sites tested showed some improvement, and I also >didn't see any especially noteworthy regressions. Thanks for testing! I will merge it soon. Best, Rodrigo

Oct. 5, 2024

Re: Issue 187: "Configure should stop when no compiler is found"

by Rodrigo Arias

Hi Alex, Thanks for the suggestions. On Sat, Oct 05, 2024 at 01:28:57PM +0200, a1ex(a)dismail.de wrote: >Hi Rodrigo, > >Regarding this issue[1], I have found couple of ways to address this: > >The easiest way is to do AC_PROG_CXX before AC_PROG_CC: > >--- a/configure.ac Sat Oct 5 13:15:28 2024 >+++ b/configure.ac Sat Oct 5 13:15:48 2024 >@@ -124,8 +124,8 @@ if test "x$enable_ssl" != "x"; then > AC_MSG_ERROR([The flag --enable-ssl is deprecated, use --enable-tls]) > fi > >-AC_PROG_CC > AC_PROG_CXX >+AC_PROG_CC > AC_PROG_RANLIB > AC_PROG_CPP > >Which now errors out when no working compiler is found: >'configure: error: C++ compiler cannot create executables' I think it relies on a detail of the implementation, so I don't think we can trust it to always work. I cannot believe that in 2024 it is still not considered a bug... >Alternately, we can add a check to verify if the complier actually >exists, and if not, exit. Something like this: > >dnl ------------------------------------------------------------ >dnl AC_PROG_CXX will set CXX=g++ even if it finds no useable C++ >dnl compiler, so we have to check whether the program named by >dnl CXX exists. >dnl ------------------------------------------------------------ >dnl >AC_PROG_CXX >AC_PATH_PROG(REAL_CXX, $CXX) >HAVE_CXX=no >if test -n "$REAL_CXX"; then > HAVE_CXX=yes >fi >if HAVE_CXX="no"; then > echo "Error: No working C++ compiler found!" > exit 1 > fi This is probably better, but even if the C++ compiler is in the path, it doesn't mean it actually works. I think something like this may be more strict, as we test that it can compile a simple C++ program: diff --git a/configure.ac b/configure.ac index 5c308cb7..7d156b81 100644 --- a/configure.ac +++ b/configure.ac @@ -129,6 +129,19 @@ AC_PROG_CXX AC_PROG_RANLIB AC_PROG_CPP +AC_MSG_CHECKING(if C++ compiler '$CXX' works) +AC_LANG_PUSH([C++]) +AC_COMPILE_IFELSE([AC_LANG_PROGRAM([[ + #ifndef __cplusplus + #error "No C++ support, AC_PROG_CXX failed" + #endif + ]])], + [AC_MSG_RESULT([yes])], + [AC_MSG_RESULT(no) + AC_MSG_FAILURE([C++ compiler doesn't work])] +) +AC_LANG_POP([C++]) + dnl ---------------------------- dnl Check our char and int types dnl ---------------------------- Outputs this when configured with the non-existent iguana C++ compiler: $ ../configure CXX=iguana ... checking whether the compiler supports GNU C++... no checking whether iguana accepts -g... no checking for iguana option to enable C++11 features... unsupported checking for iguana option to enable C++98 features... unsupported checking dependency style of iguana... none checking for ranlib... ranlib checking how to run the C preprocessor... gcc -E checking if C++ compiler 'iguana' works... no configure: error: in '/home/ram/dev/dillo/git/build': configure: error: C++ compiler doesn't work See 'config.log' for more details Best, Rodrigo.

Oct. 5, 2024

Re: CSS extra units

by a1ex＠dismail.de

Hi, Rodrigo Arias <rodarima(a)gmail.com> wrote: > This PR adds support for extra CSS units (ch, rem, vw, vh, vmin and > vmax) from the mobilized fork: > > https://github.com/dillo-browser/dillo/pull/264 > > I tested it with a few sites, and overall it improves the rending > layout, but it would be nice if you can test it with more sites > before it gets merged. After a/b testing this with a fair amount of sites, I think it's a good PR. Maybe 10-20% of sites tested showed some improvement, and I also didn't see any especially noteworthy regressions. Regards, Alex

Oct. 5, 2024

Issue 187: "Configure should stop when no compiler is found"

by a1ex＠dismail.de

Hi Rodrigo, Regarding this issue[1], I have found couple of ways to address this: The easiest way is to do AC_PROG_CXX before AC_PROG_CC: --- a/configure.ac Sat Oct 5 13:15:28 2024 +++ b/configure.ac Sat Oct 5 13:15:48 2024 @@ -124,8 +124,8 @@ if test "x$enable_ssl" != "x"; then AC_MSG_ERROR([The flag --enable-ssl is deprecated, use --enable-tls]) fi -AC_PROG_CC AC_PROG_CXX +AC_PROG_CC AC_PROG_RANLIB AC_PROG_CPP Which now errors out when no working compiler is found: 'configure: error: C++ compiler cannot create executables' Alternately, we can add a check to verify if the complier actually exists, and if not, exit. Something like this: dnl ------------------------------------------------------------ dnl AC_PROG_CXX will set CXX=g++ even if it finds no useable C++ dnl compiler, so we have to check whether the program named by dnl CXX exists. dnl ------------------------------------------------------------ dnl AC_PROG_CXX AC_PATH_PROG(REAL_CXX, $CXX) HAVE_CXX=no if test -n "$REAL_CXX"; then HAVE_CXX=yes fi if HAVE_CXX="no"; then echo "Error: No working C++ compiler found!" exit 1 fi If either of these solutions work for you, I can create a pull request. [1] https://github.com/dillo-browser/dillo/issues/187 Regards, Alex

Oct. 5, 2024

CSS extra units

by Rodrigo Arias

Hi, This PR adds support for extra CSS units (ch, rem, vw, vh, vmin and vmax) from the mobilized fork: https://github.com/dillo-browser/dillo/pull/264 I tested it with a few sites, and overall it improves the rending layout, but it would be nice if you can test it with more sites before it gets merged. Best, Rodrigo.

Oct. 3, 2024

Re: POSIX build

by Rodrigo Arias

On Thu, Oct 03, 2024 at 01:08:46PM +0200, a1ex(a)dismail.de wrote: >Done: >https://github.com/dillo-browser/dillo/pull/265 Thanks!, merged in: To github.com:dillo-browser/dillo.git e7fee996..9880c1ba master -> master Best, Rodrigo.

Oct. 3, 2024

Re: POSIX build

by a1ex＠dismail.de

Rodrigo Arias <rodarima(a)gmail.com> wrote: > Hi Alex, > > On Wed, Oct 02, 2024 at 11:44:53AM +0200, a1ex(a)dismail.de wrote: > >Hi, > > > >Regarding the recent github discussion about build systems and POSIX: > > > >Just in case nobody else gets around to it, the attached patch > >changes the build to be POSIX compliant by using 'command -v' > >instead of 'which'. > > Thanks, can you open a PR on GitHub and CC @Kangie? Done: https://github.com/dillo-browser/dillo/pull/265

Oct. 3, 2024

Re: POSIX build

by Rodrigo Arias

Hi Alex, On Wed, Oct 02, 2024 at 11:44:53AM +0200, a1ex(a)dismail.de wrote: >Hi, > >Regarding the recent github discussion about build systems and POSIX: > >Just in case nobody else gets around to it, the attached patch changes >the build to be POSIX compliant by using 'command -v' instead of >'which'. Thanks, can you open a PR on GitHub and CC @Kangie? Best, Rodrigo.

Oct. 2, 2024

POSIX build

by a1ex＠dismail.de

Hi, Regarding the recent github discussion about build systems and POSIX: Just in case nobody else gets around to it, the attached patch changes the build to be POSIX compliant by using 'command -v' instead of 'which'. -Alex

Oct. 2, 2024