Created attachment 645008 [details] [diff] [review]
comm-central: switch nsnull to nullptr
Bug 626472 is dropping nsnull in favour of nullptr.
I talked to Ehsan, and I think the plan is to land the definition of nullptr soon, and once it's available in mozilla-central, push our patch to try, and test building it using both clang and gcc.
I'll probably get our try server to apply the removal of nsnull against our mozilla-central subdirectory, for added assurance that we still build alright.
Part 1 from Bug 626472 has landed in mozilla-central.
As per our plan, I'm going to push the Attachment 645008 [details] [diff] to try to see if we get green builds.
FWIW, nullptr is now on mozilla-central, so you guys can proceed with this whenever you want.
Ok, here are our try results:
Mostly green, but not perfect. The build failures on Windows seem to be unrelated. The bloattest crashes on Linux, and the crashes caused by Mozmill are more disturbing.
Cc'ing Irving, who's awesome at diagnosing crashes and stuff.
I'll start by poking at the Mozmill crashes.
So the crashes on mailbloat seem to also be happening on trunk (sigh, they just showed up this morning), so that appears to be unrelated.
FWIW it will be really surprising if this results in crashes etc once you get green builds...
Yeah, it turns out that other stuff landed on mozilla-central that broke us in new and wonderful ways.
I think we're getting close to sorting it all out. Anyhow, those reds on the try build can be safely ignored, and I think we're going to survive the landing.
This spills outside of mailnews, and affects everything under comm-central.
Created attachment 647163 [details] [diff] [review]
Switch nsnull to nullptr
Here's an up to date conversion patch
Comment on attachment 647163 [details] [diff] [review]
Switch nsnull to nullptr
Review of attachment 647163 [details] [diff] [review]:
r=philipp for calendar parts
My try build looks good (well, good enough, considering the current bustedness of the tree): https://tbpl.mozilla.org/?tree=Thunderbird-Try&rev=c2c1e3d40065
Fallen has confirmed that the Calendar bits look sane. I've leafed through the diff, and I don't see anything crazy. What's the protocol for a patch this big?
Should we get a swarm of people to ensure that this thing is OK, or should we just land it?
On m-c, we just did sed -i 's/\bnsnull\b/nullptr/g' and ehsan gave me r+ for whatever that produced without manual review. It's really unlikely this could cause problems, since nsnull is currently defined to be nullptr in m-c!
Created attachment 648339 [details] [diff] [review]
Unbitrotted patch (checked in)
Re-ran sed script to un-bitrot, and committed to comm-central as