Horton

Thread

Commits

  1. Fix check for trapping exit() calls in libpq

  2. Provide error hint on exit() check when building libpq

  1. [PATCH] Print error when libpq-refs-stamp fails

    Rachel Heaton <rachelmheaton@gmail.com> — 2021-09-27T22:55:02Z

    Hello,

While developing I got this error and it was difficult to figure out what
was going on.

Thanks to Jacob, I was able to learn the context of the failure, so we
created this small patch.


The text of the error message, of course, is up for debate, but hopefully
this will make it more clear to others.


Thank you,
Rachel Heaton

  2. Re: [PATCH] Print error when libpq-refs-stamp fails

    Daniel Gustafsson <daniel@yesql.se> — 2021-09-28T13:14:04Z

    > On 28 Sep 2021, at 00:55, Rachel Heaton <rachelmheaton@gmail.com> wrote:

> While developing I got this error and it was difficult to figure out what was going on. 
> 
> Thanks to Jacob, I was able to learn the context of the failure, so we created this small patch. 

I can see that, and I think this patch makes sense even though we don't have
much of a precedent for outputting informational messages from Makefiles.

> The text of the error message, of course, is up for debate, but hopefully this will make it more clear to others. 

+	echo 'libpq must not call exit'; exit 1; \

Since it's not actually libpq which calls exit (no such patch would ever be
committed), I think it would be clearer to indicate that a library linked to is
the culprit.  How about something like "libpq must not be linked against any
library calling exit"?

--
Daniel Gustafsson		https://vmware.com/

  3. Re: [PATCH] Print error when libpq-refs-stamp fails

    Rachel Heaton <rachelmheaton@gmail.com> — 2021-09-28T15:52:59Z

    On Tue, Sep 28, 2021 at 6:14 AM Daniel Gustafsson <daniel@yesql.se> wrote:

>
> Since it's not actually libpq which calls exit (no such patch would ever be
> committed), I think it would be clearer to indicate that a library linked to is
> the culprit.  How about something like "libpq must not be linked against any
> library calling exit"?
>

Excellent update to the error message. Patch attached.

  4. Re: [PATCH] Print error when libpq-refs-stamp fails

    Daniel Gustafsson <daniel@yesql.se> — 2021-10-04T12:35:22Z

    > On 28 Sep 2021, at 17:52, Rachel Heaton <rachelmheaton@gmail.com> wrote:

> Patch attached.

I tweaked the error message a little bit and pushed to master. Thanks!

--
Daniel Gustafsson		https://vmware.com/

  5. Re: [PATCH] Print error when libpq-refs-stamp fails

    Anton Voloshin <a.voloshin@postgrespro.ru> — 2021-10-04T16:40:02Z

    Hello,

On 28/09/2021 05:55, Rachel Heaton wrote:
> Hello,
> 
> While developing I got this error and it was difficult to figure out 
> what was going on.
> 
> Thanks to Jacob, I was able to learn the context of the failure, so we 
> created this small patch.

-	! nm -A -u $< 2>/dev/null | grep -v __cxa_atexit | grep exit
+	@if nm -a -u $< 2>/dev/null | grep -v __cxa_atexit | grep exit; then \
+		echo 'libpq must not be linked against any library calling exit'; 
exit 1; \
+	fi

Could you please confirm that the change from -A to -a in nm arguments 
in this patch is intentional?

-- 
Anton Voloshin
Postgres Professional, The Russian Postgres Company
https://postgrespro.ru

  6. Re: [PATCH] Print error when libpq-refs-stamp fails

    Jacob Champion <pchampion@vmware.com> — 2021-10-04T17:02:11Z

    On Mon, 2021-10-04 at 23:40 +0700, Anton Voloshin wrote:
> 
> Could you please confirm that the change from -A to -a in nm arguments 
> in this patch is intentional?

That was not intended by us, thank you for the catch! A stray
lowercasing in vim, perhaps.

--Jacob

  7. Re: [PATCH] Print error when libpq-refs-stamp fails

    Daniel Gustafsson <daniel@yesql.se> — 2021-10-04T17:21:25Z

    
> On 4 Oct 2021, at 19:02, Jacob Champion <pchampion@vmware.com> wrote:
> 
> On Mon, 2021-10-04 at 23:40 +0700, Anton Voloshin wrote:
>> 
>> Could you please confirm that the change from -A to -a in nm arguments 
>> in this patch is intentional?
> 
> That was not intended by us, thank you for the catch! A stray
> lowercasing in vim, perhaps.

Hmm, I will take care of this shortly.

/ Daniel

  8. Re: [PATCH] Print error when libpq-refs-stamp fails

    Daniel Gustafsson <daniel@yesql.se> — 2021-10-04T18:36:54Z

    > On 4 Oct 2021, at 19:21, Daniel Gustafsson <daniel@yesql.se> wrote:
> 
>> On 4 Oct 2021, at 19:02, Jacob Champion <pchampion@vmware.com> wrote:
>> 
>> On Mon, 2021-10-04 at 23:40 +0700, Anton Voloshin wrote:
>>> 
>>> Could you please confirm that the change from -A to -a in nm arguments 
>>> in this patch is intentional?
>> 
>> That was not intended by us, thank you for the catch! A stray
>> lowercasing in vim, perhaps.
> 
> Hmm, I will take care of this shortly.

Right, so I missed this in reviewing and testing, and I know why the latter
didn't catch it.  nm -A and -a outputs the same thing *for this input* on my
Debian and macOS boxes, with the small difference that -A prefixes the line
with the name of the input file.  -a also include debugger symbols, but for
this usage it didn't alter the results.  I will go ahead and fix this, thanks
for catching it! 

--
Daniel Gustafsson		https://vmware.com/

  9. Re: [PATCH] Print error when libpq-refs-stamp fails

    Rachel Heaton <rachelmheaton@gmail.com> — 2021-10-05T00:09:29Z

    Thanks to you both!

From: Daniel Gustafsson <daniel@yesql.se>
Date: Monday, October 4, 2021 at 11:36 AM
To: Jacob Champion <pchampion@vmware.com>
Cc: Rachel Heaton <rachelmheaton@gmail.com>, pgsql-hackers@lists.postgresql.org <pgsql-hackers@lists.postgresql.org>, a.voloshin@postgrespro.ru <a.voloshin@postgrespro.ru>
Subject: Re: [PATCH] Print error when libpq-refs-stamp fails
> On 4 Oct 2021, at 19:21, Daniel Gustafsson <daniel@yesql.se> wrote:
>
>> On 4 Oct 2021, at 19:02, Jacob Champion <pchampion@vmware.com> wrote:
>>
>> On Mon, 2021-10-04 at 23:40 +0700, Anton Voloshin wrote:
>>>
>>> Could you please confirm that the change from -A to -a in nm arguments
>>> in this patch is intentional?
>>
>> That was not intended by us, thank you for the catch! A stray
>> lowercasing in vim, perhaps.
>
> Hmm, I will take care of this shortly.

Right, so I missed this in reviewing and testing, and I know why the latter
didn't catch it.  nm -A and -a outputs the same thing *for this input* on my
Debian and macOS boxes, with the small difference that -A prefixes the line
with the name of the input file.  -a also include debugger symbols, but for
this usage it didn't alter the results.  I will go ahead and fix this, thanks
for catching it!

--
Daniel Gustafsson               https://vmware.com/