Re: install doesn't work on HEAD

Поиск
Список
Период
Сортировка
От Ashutosh Sharma
Тема Re: install doesn't work on HEAD
Дата
Msg-id CAE9k0PkvQBv16Y4qPdwzZyvbOQiiPcKYz5qZ+ALf4jqDzNw9BA@mail.gmail.com
обсуждение исходный текст
Ответ на install doesn't work on HEAD  (Amit Kapila <amit.kapila16@gmail.com>)
Ответы Re: install doesn't work on HEAD  (Amit Kapila <amit.kapila16@gmail.com>)
Список pgsql-hackers
Hi,


On Tue, Jun 5, 2018 at 7:08 PM, Amit Kapila <amit.kapila16@gmail.com> wrote:
>
> Hi,
>
> On my win-7, I am facing $SUBJECT.  I am consistently getting below error:
> Copying build output files...Could not copy postgres.exe
>  at install.pl line 26.
>
> On digging, I found that it started failing with commit 3a7cc727 (Don't fall off the end of perl functions).  It
seemsthat we have added empty 'return' in all of the functions which seems to break one of the case:
 
>
> lcopy("$conf\\$pf\\$pf.$ext", "$target\\$dir\\$pf.$ext")
> || croak "Could not copy $pf.$ext\n";
>
> I think the return from lcopy implies 0 (or undef) which cause it to fail.  By reading couple of articles on net
[1][2]related to this, it seems we can't blindly return in all cases.  On changing, the lcopy as below, install appears
tobe working.
 
>
> @@ -40,7 +40,7 @@ sub lcopy
>         copy($src, $target)
>           || confess "Could not copy $src to $target\n";
> -       return;
> +       return 1;
>  }
>
> I have randomly check some of the other functions where this patch has added 'return' and those seem to be fine.
>
> Is it by any chance related Perl version or some other settings?  If not, then we should do something for it.
>
> [1] - https://perlmaven.com/how-to-return-undef-from-a-function
> [2] - http://search.cpan.org/dist/Perl-Critic/lib/Perl/Critic/Policy/Subroutines/RequireFinalReturn.pm

I am not able to reproduce this issue on HEAD. Seems like the
following git-commit fixes it,

commit 01deec5f8ae64b5120cc8c93d54fe0e19e477b02
Author: Andrew Dunstan <andrew@dunslane.net>
Date:   Mon May 28 16:44:13 2018 -0400

    Return a value from Install.pm's lcopy function

    Commit 3a7cc727c was a little over eager about adding an explicit return
    to this function, whose value is checked in most call sites. This change
    reverses that and returns the expected value explicitly. It also adds a
    check to the one call site lacking one.

-- 
With Regards,
Ashutosh Sharma
EnterpriseDB:http://www.enterprisedb.com

>
>
> --
> With Regards,
> Amit Kapila.
> EnterpriseDB: http://www.enterprisedb.com


В списке pgsql-hackers по дате отправления:

Предыдущее
От: Laurenz Albe
Дата:
Сообщение: Re: Loaded footgun open_datasync on Windows
Следующее
От: Tom Lane
Дата:
Сообщение: Re: commitfest 2018-07