[cfe-dev] [3.8 Release] Please write release notes!

Anna Zaks via cfe-dev cfe-dev at lists.llvm.org
Wed Mar 2 09:13:46 PST 2016

> On Mar 2, 2016, at 7:05 AM, Gábor Horváth <xazax.hun at gmail.com> wrote:
> On 12 February 2016 at 01:16, Hans Wennborg via cfe-dev <cfe-dev at lists.llvm.org <mailto:cfe-dev at lists.llvm.org>> wrote:
> Dear lots of people,
> The first comments on the 3.7 release expressed surprise that there
> were no changes to the X86 or ARM targets. There had of course been a
> lot of hard work and many changes, but none of it was mentioned in the
> release notes.
> Please help make the release notes more comprehensive this time. The
> notes are of course not as important as the actual code, but they do
> get read, and they are a good way of telling users about all the hard
> work you have done in the past six months.
> If you made any changes in the 3.8 time span that might be worth
> mentioning, please add them to the release notes --- it does not have
> to be big-ticket items. If you saw someone else commit something
> interesting, prod them to write a note about it.
> The 3.8-rc2 notes can be seen at [1]. To add something, commit it
> directly to the branch, or if you prefer, send me a patch or even just
> an email with some text. I would be very happy if we can get these
> into good shape by the beginning of next week.
> Some things that might be worth mentioning, in no particular order:
> (based partially on Alex Bradbury's LLVM Weekly; thanks!)
> - Intel, ARM, PowerPC, MIPS, SPARC, Hexagon, ..., target maintainers,
> please write something about your target.
> - Eric: should the new C API Changes part of the developer policy be
> pointed out? Maybe in relation to where we mention changes to the C
> API?
> - David: can you write something about the new IR for exception
> handling? Maybe just point into the ExceptionHandling doc?
> - Alex L: I've seen a lot of MIR patches but don't know the status.
> Anything to mention in the release notes?
> - Dylan: should we mention the new AVR target, or is it too early?
> - Richard: anything on the C++ front? What's the status on coroutines?
> - Jonathan and other OpenMP folks: it's no longer behind a flag; would
> be great to get some notes on this.
> - Artem: there's been a bunch of CUDA patches. Anything that should be
> mentioned in the release notes?
> - Jordan & Anna: any new checkers that should be mentioned?
> I did not see any response yet, so I cherry picked some changes:
> - There are better diagnostics for loading analyzer plugins, when the static analyzer is invoked from libtooling.
> - New checks introduced for iOS / OS X localizability issues.
> - New warning, when a nested if has the same condition (without side effects) as the outer if.
> - New checks for the ObjC lightweight generics feature.
> - New checks that utilize nullability qualifiers.
> - New check to find unsafe uses of vforked projects.
> - New check to detect excess padding in records.
> - Better modelling memcpy.
> - Better modelling messaging to nil in ObjC.
> - New API to generate error nodes.
> - Fixes to the modelling of pointer arithmetic.
> - Better handle range switch cases.
> - Better handle realloc, fix some false positives in malloc-overflow check.
> - Better model const methods.
> - Support for lambda functions.
> - Experimental feature to widen loops. When enabled, the analyzer will have greater coverage in code that has loops with high constant bounds.
> - Generate bug hash as an identifier for each report. It can be used to track bugs in evolving code.
> - Improve the modelling of static initializers.
> - Better model stack allocated blocks in Obj-C.
> - Better model nullptr_t.
> - Better model casts between integral types.
> - Various other improvements in the analyzer engine.
> Feel free to reword and pick some of those into the release notes.

Thank you Gabor,

We’ve updated the Static Analyzer notes with the following summary. Looks like there are no glaring omissions from the list you’ve provided. I chose not to list the specific improvements in the core to keep this focused and user-friendly.

+The scan-build and scan-view tools will now be installed with clang. Use these
+tools to run the static analyzer on projects and view the produced results.
+Static analysis of C++ lambdas has been greatly improved, including
+interprocedural analysis of lambda applications.
+Several new checks were added:
+- The analyzer now checks for misuse of ``vfork()``.
+- The analyzer can now detect excessively-padded structs. This check can be
+  enabled by passing the following command to scan-build:
+  ``-enable-checker optin.performance.Padding``.
+- The checks to detect misuse of ``_Nonnull`` type qualifiers as well as checks
+  to detect misuse of Objective-C generics were added.
+- The analyzer now has opt in checks to detect localization errors in Coca
+  applications. The checks warn about uses of non-localized ``NSStrings``
+  passed to UI methods expecting localized strings and on ``NSLocalizedString``
+  macros that are missing the comment argument. These can be enabled by passing
+  the following command to scan-build:
+  ``-enable-checker optin.osx.cocoa.localizability``.

> - Chih-Hung: should the notes mention the emutls-style TLS model?
> - Lang: you did a lot of work on the Kalidoscope tutorial. Maybe we
> should mention that?
> .. tell me what I missed :-)
> Thanks,
> Hans
>  [1]. http://llvm.org/pre-releases/3.8.0/#rc2 <http://llvm.org/pre-releases/3.8.0/#rc2>.
> _______________________________________________
> cfe-dev mailing list
> cfe-dev at lists.llvm.org <mailto:cfe-dev at lists.llvm.org>
> http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-dev <http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-dev>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20160302/dcfc202b/attachment.html>

More information about the cfe-dev mailing list