Re: [PATCH 8/8] Use require with is_remote

[Pedro Alves <pedro@palves.net>]

On 2023-01-22 9:55 p.m., Tom Tromey wrote:
> --- a/gdb/testsuite/gdb.base/fullpath-expand.exp
> +++ b/gdb/testsuite/gdb.base/fullpath-expand.exp
> @@ -15,10 +15,7 @@
>  
>  standard_testfile .c fullpath-expand-func.c
>  
> -if [is_remote host] {
> -    unsupported "compiling on a remote host does not support a filename with directory."
> -    return 0
> -}
> +require {!is_remote host}

IWBN to preserve "compiling on a remote host does not support a filename with directory."
as a comment.

> --- a/gdb/testsuite/gdb.base/gdbhistsize-history.exp
> +++ b/gdb/testsuite/gdb.base/gdbhistsize-history.exp
> @@ -21,10 +21,7 @@
>  # We cannot expect remote hosts to see environment variables set on the
>  # local machine.
>  
> -if { [is_remote host] } {
> -    unsupported "can't set environment variables on remote host"
> -    return -1
> -}
> +require {!is_remote host}

Similarly here.

> diff --git a/gdb/testsuite/gdb.base/solib-corrupted.exp b/gdb/testsuite/gdb.base/solib-corrupted.exp
> index 644c8ece11f..a2fbaa6ac40 100644
> --- a/gdb/testsuite/gdb.base/solib-corrupted.exp
> +++ b/gdb/testsuite/gdb.base/solib-corrupted.exp
> @@ -15,11 +15,7 @@
>  
>  require allow_shlib_tests
>  
> -if {[is_remote target]} {
> -    # gdbserver prints the warning message but expect is parsing only the GDB
> -    # output, not the gdbserver output.
> -    return 0
> -}
> +require {!is_remote target}

Ditto, comment should be preserved, IMO.  

There are a few more, I've not pointed at every one.

Basically, these help us understand what particular issue with is_remote boards causes
the test to be skipped.

The cases where the messages were just "doesn't work on remote host" such as:

 > -if {[is_remote host]} {
 > -    untested "does not work on remote host"
 > -    return 0
 > -}
 > +require {!is_remote host}

are useless and should not be preserved of course.