Hi Alexandre,
> On Jun 9, 2020, Rainer Orth <r...@cebitec.uni-bielefeld.de> wrote:
>
>> this is wrong unfortunately: braces are the Tcl equivalent of single
>> quotes so you're setting skip_lto to the string inside.
>
> Aah, thanks. So when $skip_lto is expanded in the ifs, the whole thing
> is evaluated, and that's why it works anyway?
but it didn't work: when I tried the patch on i386-pc-solaris2.11 with
as/ld (i.e. no linker plugin}, the lto tests of outputs.exp were still
run, and check_linker_plugin_available wasn't even run at all, what
quite confused me at first. As I'd mentioned in the PR, you can easily
test an equivalent configuration with gld when configuring gcc with
--disable-lto-plugin.
>> While this worked for me on i386-pc-solaris2.11 (both with ld, thus
>> without the lto plugin, and with gld and the plugin), I wonder if
>> there's not a better way than just skipping the lto tests,
>
> My initial focus was on relieving the pain by removing the symptoms, so
> I can then work on the improvements with lower urgency.
Fully understood: as I'd mentioned, that's mine as well. We already
accumulate all too many testsuite failures on master...
> Here's what I'd like to check in for now:
>
> [PR95416] outputs.exp: skip lto tests when not using linker plugin
>
> From: Alexandre Oliva <ol...@adacore.com>
>
> When the linker plugin is not available, dump outputs in lto runs are
> different from those outputs.exp tests expect, so skip them for now.
>
> Regstrapped on x86_64-linux-gnu. Ok to install?
>
>
> for gcc/testsuite/ChangeLog
>
> * outputs.exp (skip_lto): Set when missing the linker plugin.
Ok, thanks.
Rainer
--
-----------------------------------------------------------------------------
Rainer Orth, Center for Biotechnology, Bielefeld University
