Revert "remove optabgen.d"

[Geod24]

Reverts #10221

Let's see what SemaphoreCI says.
(I wish Github opened revert branches in forks...)

[dlang-bot]

Thanks for your pull request, @Geod24!

Bugzilla references

Your PR doesn't reference any Bugzilla issue.

If your PR contains non-trivial changes, please reference a Bugzilla issue or create a manual changelog.

Testing this PR locally

If you don't have a local development environment setup, you can use Digger to test this PR:

dub fetch digger
dub run digger -- build "master + dmd#10238"

[marler8997]

What's the reason for the revert?

[rainers]

What's the reason for the revert?

master failing to build on semaphoreci: https://semaphoreci.com/dlang/dmd-2/branches/master

[Geod24]

The PR broke SemaphoreCI. Master is currently red: https://semaphoreci.com/dlang/dmd-2
But this PR fixes it.
CC @thewilsonator

[marler8997]

Shoot, now I'm going to have to rebase #10217 after already rebasing it from #10221. Then if this gets fixed and re-merged, then I"ll have to rebase/rework again. Is it possible to try to fix the failure before reverting?

[Geod24]

You can just use git reflog (and Github might still show the commits).
In any case, having the reverter fix the bug is a bad trend: it shift the burden of work to the person that finds the bug, not the one that creates the bug.

[marler8997]

You can just use git reflog (and Github might still show the commits).

Sure but I've made changes since the rebase so my original commit is no longer valid. It's not a big deal, I'm just letting you know that this is causing extra work.

In any case, having the reverter fix the bug is a bad trend: it shift the burden of work to the person that finds the bug, not the one that creates the bug.

I'm not worried about trends. You have every right to revert the commit, but that doesn't mean we can't try to fix it first either. You don't have a responsibility to fix it of course, I was just asking if it could be easily done. If not then by all means go ahead with the revert (not that you need my permission or anything).

[marler8997]

Ping @WalterBright, do you want a chance to come up with a quick fix before the change is reverted? Or should we go ahead with the revert now and fix this later?

[marler8997]

Interesting, semaphore is currently passing with my PR, which has been rebased with the "remove optabgen": https://semaphoreci.com/dlang/dmd-2/branches/pull-request-10212/builds/22

[marler8997]

Well I couldn't figure out the cause in the few minutes I spent, so don't hold back on my account. I'm off to bed.

[rainers]

I don't see why removing an usused generator can produce different binaries, but I suspect we hit some threshold producing different code due to something like

dmd/src/dmd/backend/go.d

Line 351 in 08c385c

if (go.changes && go.mfoptim & MFloop && (os_clock() - starttime) < 30 * CLOCKS_PER_SEC)

in the optimizer.

[rainers]

Maybe #10240 helps.

[rainers]

Maybe #10240 helps.

it didn't but #10242 seems to.

[Geod24]

Thanks @rainers !