[Lldb-commits] [lldb] 6c0cd56 - [lldb] Make `process connect` behave the same in sync and async mode.

[Jonas Devlieghere via lldb-commits]

I'm OOO this week but I've deleted the test until I can write a better one
next week.

On Mon, Nov 23, 2020 at 5:21 AM Pavel Labath <pavel at labath.sk> wrote:

> On 14/11/2020 02:39, Jonas Devlieghere via lldb-commits wrote:
> > +# RUN: echo -e 'platform select remote-gdb-server\nprocess connect
> connect://localhost:4321' | %lldb 2>&1 | FileCheck %s
> > +
> > +# CHECK: Platform: remote-gdb-server
> > +# CHECK: Connected: no
> > +# CHECK: error: Failed to connect port
>
> This is a bad test. You can't assume that the host machine will not have
> port 4321 open. If it does, for some reason, the test will fail.
> Moreover, if that port happens to belong to some other test (not likely,
> as we normally use higher port numbers, but possible), the spurious
> connection cause *that* test to fail as well...
>
> (We've observed flakyness in this test internally. While I don't have
> conclusive proof that this is the cause, it's a pretty good bet that it
> is.)
>
> The right way to handle this would be to first claim a port, and then
> have lldb server connect to that port. (most likely requires writing the
> test in python.)
>
> pl
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/lldb-commits/attachments/20201123/8449f2cf/attachment.html>