Bug 1943056

Are you referring to https://copr.fedorainfracloud.org/coprs/odubaj/autoconf-2.70/build/2095799/ with your report? If so, then I would say this is not autoconf related, because it also happens on Fedora Rawhide directly and looks like this:

--- 8< ---
[…]
make[2]: Entering directory '/builddir/build/BUILD/partclone-0.3.17/fail-mbr'
sh compile-mbr.sh
Checking the file:
files fail-mbr.bin and fail-mbr.bin.orig differ significantly:
2c2
< fail-mbr.bin:     file format binary
---
> fail-mbr.bin.orig:     file format binary
8,28c8,36
<    0:	04 00                	add    $0x0,%al
<    2:	00 00                	add    %al,(%bx,%si)
<    4:	20 00                	and    %al,(%bx,%si)
<    6:	00 00                	add    %al,(%bx,%si)
<    8:	05 00 00             	add    $0x0,%ax
<    b:	00 47 4e             	add    %al,0x4e(%bx)
<    e:	55                   	push   %bp
<    f:	00 02                	add    %al,(%bp,%si)
<   11:	00 01                	add    %al,(%bx,%di)
<   13:	c0 04 00             	rolb   $0x0,(%si)
< 	...
<   1e:	00 00                	add    %al,(%bx,%si)
<   20:	01 00                	add    %ax,(%bx,%si)
<   22:	01 c0                	add    %ax,%ax
<   24:	04 00                	add    $0x0,%al
<   26:	00 00                	add    %al,(%bx,%si)
<   28:	01 00                	add    %ax,(%bx,%si)
<   2a:	00 00                	add    %al,(%bx,%si)
<   2c:	00 00                	add    %al,(%bx,%si)
<   2e:	00 00                	add    %al,(%bx,%si)
<   30:	00 74 fc             	add    %dh,-0x4(%si)
---
>    0:	eb 0e                	jmp    0x10
>    2:	4f                   	dec    %di
>    3:	43                   	inc    %bx
>    4:	53                   	push   %bx
>    5:	20 44 75             	and    %al,0x75(%si)
>    8:	6d                   	insw   (%dx),%es:(%di)
>    9:	6d                   	insw   (%dx),%es:(%di)
>    a:	79 20                	jns    0x2c
>    c:	4d                   	dec    %bp
>    d:	42                   	inc    %dx
>    e:	52                   	push   %dx
>    f:	90                   	nop
>   10:	33 c0                	xor    %ax,%ax
>   12:	50                   	push   %ax
>   13:	1f                   	pop    %ds
>   14:	be 33 7c             	mov    $0x7c33,%si
>   17:	ac                   	lods   %ds:(%si),%al
>   18:	3c 00                	cmp    $0x0,%al
>   1a:	74 0c                	je     0x28
>   1c:	bb 07 00             	mov    $0x7,%bx
>   1f:	b9 01 00             	mov    $0x1,%cx
>   22:	b4 0e                	mov    $0xe,%ah
>   24:	cd 10                	int    $0x10
>   26:	eb ef                	jmp    0x17
>   28:	b4 02                	mov    $0x2,%ah
>   2a:	ba 00 01             	mov    $0x100,%dx
>   2d:	cd 10                	int    $0x10
>   2f:	3c 00                	cmp    $0x0,%al
>   31:	74 fc                	je     0x2f
make[2]: Leaving directory '/builddir/build/BUILD/partclone-0.3.17/fail-mbr'
compile-mbr.sh: line 2: dpkg-architecture: command not found
compile-mbr.sh: line 2: dpkg-architecture: command not found
make[2]: *** [Makefile:504: fail-mbr.bin] Error 1
make[1]: *** [Makefile:401: all-recursive] Error 1
make: *** [Makefile:342: all] Error 2
--- 8< ---

This build failure is IMHO, because some parts of the build-chain (gcc? annobin?) changed too much to still provide "reproducible" builds, that could be somehow matched fuzzy. But I would be grateful if somebody with low-level experience here could step in anyway :)

This bug was automatically created according to FTBFS with autoconf-2.71. No other issues were investigated.

For now, I've filed https://github.com/Thomas-Tsai/partclone/issues/159 at upstream - with the hope that somebody can point out whether there is a regression or whether e.g. the reference files in partclone only need to be updated.

Gentle ping.

Gentle pong. As mentioned before this FTBFS is not related to autoconf 2.7.1 and as per comment #3, there has been no upstream reaction so far. Do you have any other suggestion rather to disable the build-time comparision?

Probably not, thanks for staying us informed!

Hi, are there any updates?

Thanks.

Still no upstream response, but still not related to autoconf 2.7.1. The workaround is to disable upstream's check comparision, which is more likely to be related to GCC or other buildchain changes. So nothing which would hold up autoconf 2.7.1 from entering Rawhide.