On Fri, Nov 6, 2026, at 8:52 PM, Pietro Monteiro via Gcc wrote: > On Fri, Nov 6, 2026, at 2:46 PM, Eric Gallager via Gcc wrote: >> On Thu, Nov 5, 2026 at 9:55 PM Pietro Monteiro >> <[email protected]> wrote: >>> >>> Suspicious Content Warning: GW's email security system has determined that >>> the message below may be a potential email security threat. This message >>> includes content that is suspicious in nature and may indicate an attempt >>> to gather personal information. Proceed with caution if replying to or >>> following any links. Remember that GW will never ask you for your personal >>> information in an email. If you agree that the message below appears to be >>> suspicious, please forward it to [email protected]. >>> >>> This warning is provided to you as a service from GW's Division of >>> Information Technology. If you have any questions about the validity of a >>> link or a message you receive, contact the IT Support Center at >>> 202-994-GWIT(4948), [email protected] or https://it.gwu.edu/ >>> >>> ------------ Suspicious threat disclaimer ends here ------------ >>> >>> On Thu, Nov 5, 2026, at 12:19 PM, Eric Gallager via Gcc wrote: >>> > On Thu, Nov 5, 2026 at 12:02 PM Pietro Monteiro via Gcc <[email protected]> >>> > wrote: >>> >> >>> >> I decided to try upgrading autoconf and automake to the latest >>> >> available versions, respectively 2.73 and 1.19. I put the result in a >>> >> branch on the forge (https://forge.sourceware.org/pietro/gcc.git >>> >> branch upgrade-autotools) and I'm doing some basic testing. The patch >>> >> at the tip of that branch depends on some patches I've sent >>> >> recently[1,2,3] and a patch I haven't sent yet hat does the same as >>> >> patch [3] but for libiberty. >>> >> >>> >> Before I go any further on this and start looking into upgrading >>> >> binutils-gdb tree too I wanted to ask: >>> >> >>> >> 1) Is this something we want to do at this moment in the release cycle? >>> >> >>> >> 2) If so, are the versions I picked ok? I was told developers install >>> >> autotools from source to avoid distro patches, so I went for latest. >>> >> >>> > >>> > There is a bug with autoconf 2.73 that I would like to see fixed >>> > before upgrading: https://savannah.gnu.org/support/?111401 >>> > (Note that Savannah has been throwing 502 Bad Gateway errors on me >>> > lately, so I can't confirm the latest status of it at the moment) >>> >>> I got the page to load after a few attempts. The bug title is: >>> >>> autoconf 2.73 -std=gnu23 spilling from CC into CC_FOR_BUILD >>> >>> >>> And it seems to be about the autoconf-archive macro >>> AX_PROG_CC_FOR_BUILD, which we don't use. >> >> OK so it seems like I linked the wrong bug due to Savannah erroring on >> me; I actually meant this one: >> https://savannah.gnu.org/support/?111400 > > I tested the patch on the bug report and it fixes the issue: > https://codeberg.org/sleepy4727/autoconf-2.73-test > > I'll add it to my branch and regenerate all files.
I removed the generated files from the commits on my branch and opened a PR on the forge: https://forge.sourceware.org/gcc/gcc/pulls/272 This way it’s easier to see what I’ve done compared to what autoconf/automake are doing. And it makes it possible to send the patches to gcc-patches. I created a branch with all the generated files called upgrade-autotools-full to make it easier to test it. https://forge.sourceware.org/pietro/gcc/src/branch/upgrade-autotools-full pietro >>> >> 3) What kind of testing should be done? Are other people able to help >>> >> here, specially for targets that don't use the GNU toolchain? >>> >> >>> >> Thanks, >>> >> Pietro >>> >> >>> >> >>> >> [1] >>> >> https://inbox.sourceware.org/gcc-patches/[email protected]/t/#u >>> >> [2] >>> >> https://inbox.sourceware.org/gcc-patches/[email protected]/t/#u >>> >> [3] >>> >> https://inbox.sourceware.org/gcc-patches/[email protected]/t/#u
