On Wed, 2016-02-24 at 17:43 +0100, Laszlo Ersek wrote: > On 02/24/16 17:26, David Woodhouse wrote: > > On Wed, 2016-02-24 at 15:46 +0100, Laszlo Ersek wrote: > > >> Interestingly, the failure reproduces even when I build OVMF at your > >> commit a35e4359d; with identical symptoms. > > > > Also useful to know; thanks. > > The first commit that breaks it is: > > commit 6a48e82abc5174e4810e388fdf0fc67564dabb03 > Author: David Woodhouse <[email protected]> > Date: Wed Aug 12 12:54:28 2015 +0100 > > CryptoPkg/OpensslLib: Update OpenSSL patch
Haha, well that serves me right. But it gives me an excuse to rant about bad engineering practice without picking on some *other* poor sod and making them the victim, so here goes... :) In the process of identifying the individual changes we had against OpenSSL 1.0.2, coming up with acceptable alternatives to them which can be submitted to OpenSSL 1,1, and then backporting the resulting commits from OpenSSL 1.1 back to the 1.0.2 branch... I have broken at least one of them :) The reason I say it serves me right is because I should never have changed *all* of those in one big "Update OpenSSL patch" commit in my EDK2 tree in the first place. There should have been a sequence of a dozen or so commits, each of which changed just *one* part of the EDKII_openssl-1.0.2f patch to match the backported upstream version of the same change. It's just as well it was found now. If the breakage was more esoteric and was only found later, and bisected to this commit, someone would have been cursing me. I freely admit I was being lazy, because EDK2 doesn't normally follow sane engineering practices so I thought I'd get away with it. And because decent engineering practices would have meant we didn't have all those different changes lumped together into a single EDKII_openssl-1.0.2f.patch in the *first* place. But I'm just doing a drive-by here. Anyone who is *habitually* working on EDK2, if you were prepared to ack this patch as it was and not insist that I split it up — you need to rethink your source management practices. Small commits, changing *one* thing, such that the tree builds and works at each stage. No exceptions. Not even for me :) -- dwmw2
smime.p7s
Description: S/MIME cryptographic signature
_______________________________________________ edk2-devel mailing list [email protected] https://lists.01.org/mailman/listinfo/edk2-devel
