Re: MSVC installs too much stuff? - Mailing list pgsql-hackers

From Michael Paquier
Subject Re: MSVC installs too much stuff?
Date
Msg-id 20200131052657.GC2631@paquier.xyz
Whole thread Raw
In response to MSVC installs too much stuff?  (Andrew Dunstan <andrew.dunstan@2ndquadrant.com>)
Responses Re: MSVC installs too much stuff?  (Craig Ringer <craig@2ndquadrant.com>)
List pgsql-hackers
On Fri, Jan 31, 2020 at 12:47:29PM +1030, Andrew Dunstan wrote:
> When I was working on the test_json stuff yesterday, I noticed that
> there are some unexpected (by me at least) things installed when we do
> an MSVC build:
>
> $ ls -l bin| egrep 'regress|isolation'
> -rwxr-xr-x 1 pgrunner None   72192 Jan 30 07:51 isolationtester.exe
> -rwxr-xr-x 1 pgrunner None  112640 Jan 30 07:51 pg_isolation_regress.exe
> -rwxr-xr-x 1 pgrunner None  112128 Jan 30 07:51 pg_regress.exe
> -rwxr-xr-x 1 pgrunner None  112640 Jan 30 07:51 pg_regress_ecpg.exe
>
> This is made all the more obscure by the fact that the install script
> doesn't tell you exactly what it's installing, unlike the "make"
> driven install.  There could well be other things that are installed
> that shouldn't be.

+1.  Looking at vcregress.pl, all four are always invoked from the
root of the build folder.

> So I think we need to do several things:
>
> . make the install script more verbose
> . work out how to ensure the things above (and test_json when we add
> it) are not installed.
> . check that nothing else is installed that shouldn't be.

Hmm.  It seems to me that an exclusion list with patterns to match
should be enough in Install.pm.  Having only one code path for the
filtering would be nice, which means merging CopyFiles and
CopySetOfFiles.
--
Michael

Attachment

pgsql-hackers by date:

Previous
From: Kyotaro Horiguchi
Date:
Subject: Re: Proposal: Add more compile-time asserts to exposeinconsistencies.
Next
From: Kyotaro Horiguchi
Date:
Subject: Re: standby apply lag on inactive servers