Activity
From 03/10/2026 to 03/16/2026
Today
- 02:49 AM Revision 2fc3013e (git): Bump the github-actions group across 1 directory with 3 updates
- Bumps the github-actions group with 3 updates in the / directory: [ruby/setup-ruby](https://github.com/ruby/setup-ruby), [actions-rust-lang/setup-rust-toolchain](https://github.com/actions-rust-lang/setup-rust-toolchain) and [taiki-e/ins...
-
02:29 AM Revision a0fc03ef (git): [ruby/rubygems] Fix ruby/ruby detection path in bundler spec_helper
- The previous path ../../../lib/bundler/bundler.gemspec resolved outside
the repository root in both ruby/ruby and rubygems repos, so the guard
never triggered. Fix to ../../lib/bundler/bundler.gemspec which correctly
detects the ruby/rub... -
- Skip SimpleCov setup when running inside the ruby/ruby repository to
avoid interference. Detection uses GEM_COMMAND env var for rubygems
and bundler.gemspec presence for bundler.
https://github.com/ruby/rubygems/commit/48805d47f6
Co-Au... -
- Configure SimpleCov in spec_helper.rb to collect coverage data for
the bundler test suite, outputting results to the shared coverage
directory.
https://github.com/ruby/rubygems/commit/8af3fe6323
Co-Authored-By: Claude Opus 4.6 <noreply... -
- When running bundler specs with parallel_rspec, each worker process
needs a unique command_name to prevent overwriting each other's
coverage results in .resultset.json.
https://github.com/ruby/rubygems/commit/1f3ad726ac
Co-Authored-By:... -
- Add coverage_setup.rb loaded via -r flag before rubygems to enable
Coverage tracking of boot files. SimpleCov is configured in helper.rb
to collect coverage data for rubygems test suite.
https://github.com/ruby/rubygems/commit/8eed548e4... -
- https://github.com/ruby/rubygems/commit/ca2c5eda12
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> -
01:47 AM Revision 64fcb596 (git): [ruby/rubygems] Remove outdated which command isolation workaround
- https://github.com/ruby/rubygems/commit/6d4f506390
-
- https://github.com/ruby/rubygems/commit/0420c70cbf
03/15/2026
-
11:13 PM Revision 8da2777d (git): [ruby/psych] Update SnakeYAML-Engine to 3.0.1
- There are two CVEs in Guava, which is a test-scoped dependency of
this library. Guava is not shipped with the library so these CVEs
do not affect users, but the dependency may trigger security tools.
We update to avoid this false positiv... -
05:06 PM Feature #21853: Make Embedded TypedData a public API
- Eregon (Benoit Daloze) wrote in #note-5:
> also considering there is no safeguard to avoid writing to `DATA_PTR`
One idea to address this (but not the other 2 concerns) would be to raise on `DATA_PTR()` for `RUBY_TYPED_EMBEDDABLE`, a... -
- One tricky aspect about `RUBY_TYPED_EMBEDDABLE` is if in the `struct` there is a pointer to inside that `struct` then those pointers will become invalid when the object is moved.
Is there a way to handle that correctly to update such poi... -
05:06 PM Revision 9e89e5f3 (git): [DOC] Doc for Pathname.new (#16373)
-
- I'm aware what I propose doesn't solve `Can't support objects of variable width`, i.e. the case for `Thread::Backtrace`.
But AFAIK variable width is only available for core classes (not as public API), and `Thread::Backtrace` is core too... -
- Going further, `Class#safe_initialization` instead of redefining these 3 methods could just set a new `internal_alloc_func` field in RClass (only used by new/dup/clone and can never be read by anything else) + rb_undef_alloc_func().
And... -
- Thinking more about this I think there should be a protocol or an easy way to avoid the allocated-but-uninitialized state completely, which is problematic for classes defined in C but also in Ruby (though Ruby-defined classes will typica...
-
12:47 PM Revision 6297a3bb (git): [DOC] Simplify doc setup (#16393)
- * Remove stale entries from `.document` files
`doc/.document` referenced two non-existent directories (`date`, `rdoc`).
`ext/.document` referenced eight extensions that have been moved to
gems: bigdecimal, fiddle, monitor, nkf, racc, r... -
08:53 AM Feature #21953 (Open): Allow accessing unshareable objects within a Ractor-local Ruby Box
- ### Status
Currently, non-main ractors prohibit access to the following objects to prevent data races:
* Global variables
* Class variables
* Unshareable class instance variables
* Unshareable constants
### Proposal
... -
08:48 AM Misc #21877: DevMeeting-2026-03-17
- * [Feature #21861] C API: expose `ruby_xfree_sized`, `ruby_xrealloc_sized`, etc (byroot)
* C23 added `void free_sized(void *ptr, size_t old_size)`.
* It has both speed and correctness benefits.
* Several common allocators alread... -
05:42 AM Revision ca23c7c4 (git): [ruby/strscan] Do not use C99 style comment [ci skip]
- The minimum required ruby version of this library is still 2.4.
C99 has been adopted since ruby 2.7.
https://github.com/ruby/strscan/commit/c50abc95ad -
- The minimum required ruby version of this library is still 2.6.
C99 has been adopted since ruby 2.7.
https://github.com/ruby/io-console/commit/f907e5a237 -
- https://github.com/ruby/io-console/commit/d0ee6034a9
03/14/2026
-
10:19 PM Misc #21877: DevMeeting-2026-03-17
- * [Bug #21685] Unnecessary context-switching, especially bad on multi-core machines. (jpl-coconut)
* Ruby exhibits excessive context switching around fast syscalls, which compromises performance. Microbenchmarks (linked from issue and ...
03/13/2026
-
- 07:00 PM Revision 86049a98 (git): [ruby/prism] feat: add `start_line`, `end_line`, `start_column`, `end_column`, and `chop` methods for `Location`
- https://github.com/ruby/prism/commit/4258b2ad7e
Co-Authored-By: Kevin Newton <kddnewton@gmail.com> -
05:39 PM Revision ee275b41 (git): Git ignore dump_ast if you are doing an in-place build
-
- Fixed by https://github.com/ruby/ruby/commit/a70dbe1065f58a23a85e576d19e3434e6dd4bb64
-
04:35 PM Bug #21870: Regexp: Warnings when using slightly overlapping \p{...} classes
- I have submitted a patch to Onigmo with the naive strategy (A) as outlined above:
https://github.com/k-takata/Onigmo/pull/175
Contrary to my expectations, it does appear that `/[\p{Word}\p{Alpha}]/` and `/[a-fb-g]/` **do** continue... -
- https://github.com/ruby/prism/commit/0f1500ce92
-
- * TRUFFLERUBY_METRICS_REPS=5 jt metrics time --experimental-options -e0
For parsing-core:
before: 0.097 0.099 0.092 0.096
after: 0.061 0.063 0.066 0.059
* Remove extra trailing spaces by using `<%-#`.
https://github.com/ruby/pris... - 03:00 PM Revision 9246b770 (git): Bump timeout for TestGC#test_finalizer_not_run_with_vm_lock
- This test should be redesigned, but increasing the timeout
should probably be good enough for now to not see CI failures related
to it. I can see how this could timeout after 10s.
Failing test after 10s timeout:
https://ci.rvm.jp/result... -
- Prevent GC from accidentally collecting
-
- Apple's libc implementation of memmem is super slow (it is a forked
version of freebsd's that never got vectorized). Instead, we should
fall back to the rolling hash on Apple. In the attached benchmark,
I'm seeing 1.07% slower to 30.34% ... -
- 02:39 AM Revision 90f6507e (git): Bump taiki-e/install-action
- Bumps the github-actions group with 1 update in the / directory: [taiki-e/install-action](https://github.com/taiki-e/install-action).
Updates `taiki-e/install-action` from 2.68.26 to 2.68.27
- [Release notes](https://github.com/taiki-e...
03/12/2026
- 05:00 PM Revision b0754fbd (git): ZJIT: Deduplicate polymorphic send branches by receiver type (#16335)
- When lowering polymorphic `opt_send_without_block`, emit at most one branch per receiver type instead of one branch per profile bucket (`class, shape, flags`).
This deduplicates redundant `HasType`/`IfTrue` chains while preserving immed... -
04:08 PM Revision 1a0e67d3 (git): ZJIT: `::RubyVM::ZJIT.induce_side_exit!` and `induce_compile_failure!`
- Tells ZJIT to do a side exit or to fail to compile, useful testing and
for bug reports. We are picky about the syntactic form so we can tell
where the call lands early in the compiler pipeline. The `::` prefix
allows us to interpret it w... -
-
-
02:28 PM Bug #21940: Ruby::Box: `$_` returns stale value due to gvar_tbl caching
- I may have another possibly related `Ruby::Box` special-global case.
This may be a separate `$?` / `Process::Status` issue rather than the same
root cause as `$_` or `$~`, since `$?` seems to be thread-local. Still, the
symptom also... -
I found what looks like a separate `Ruby::Box` bug from the existing `require` and `LoadError` issues such as `#21760`.
This is not a `LoadError` case. I was able to reduce it to a reproducer where requiring `fiddle/import` from mul...-
- > Reducing CPU load can be useful even in tasks containing IO.
Oh absolutely, this is a huge part of my day to day work. But even for this I tend to prefer looking at a WALL report as it include both information. I see no need to ign... -
11:33 AM Feature #21950: Add a built-in CPU-time profiler
- Yes, this ticket is about sampling profilers. Sorry for the confusion.
byroot (Jean Boussier) wrote in #note-11:
> Which is an interesting choice, because maybe it's my bias, but I find that CPU-time profiling is really not as common... -
- byroot (Jean Boussier) wrote in #note-11:
> But I suspect supporting both isn't that much extra work?
I think it'd be good for this API to be able to say which one it wants (like `StackProf`'s `mode:` kwarg). -
- osyoyu (Daisuke Aritomo) wrote in #note-8:
> I don't want the API/output format to restrict what a profiler could emit. I wouldn't say such design is not doable, but careful consideration would be needed.
Careful consideration is exa... -
- > it is not about adding a sampling profiler, it is about adding a CPU time profiler that does not sample.
Well, the issue description talk of CPU-time, but it also talk about sampling (`timer_create(3)` etc). So yes, it's a CPU-time ... - 11:46 AM Revision 174b54c4 (git): [ruby/rubygems] Add a Mutex to prevent a bug on Ruby 3.2:
- - There are failing tests on Ruby 3.2 due to an expecation that
checks wheter stderr is empty. In this case, stderr outputs a
warning from Ruby "warning: loading in progress, circular require
considered harmful", but this is not tr... - 11:46 AM Revision ac3b2377 (git): [ruby/rubygems] Split the download and install process of a gem:
- - ### Problem
Bundler awaits for the dependencies of a gem to be download and installed
before it proceeds to downloading and installing the dependency itself.
This creates a bottleneck at the end of the installation process and
... -
-
- 08:25 AM Revision 235195f8 (git): [ruby/rubygems] Raise an error when building a gem that has a self reference:
- - ### Problem
A gem that has a self-reference in its dependencies would previously
get a warning during `gem build`, saying it's "discouradged".
A gem that includes a self-reference can't be updated due to bundler
filtering it o... -
03:56 AM Revision 5c81ba21 (git): [ruby/rubygems] Make SafeYAML.load an alias of safe_load
- Using Psych, load was actually more restrictive than safe_load. Using
Gem::YAMLSerializer they were identical. We might as well use the same
path for both methods.
https://github.com/ruby/rubygems/commit/d8d927f889 -
- https://github.com/ruby/rubygems/commit/51544ebfd8
-
- This had been the only user of Gem::SafeYAML.load for a long time. We
might as well be consistent with all other uses and use safe_load.
https://github.com/ruby/rubygems/commit/1b698779f5 -
- https://github.com/ruby/rubygems/commit/227df53bc5
-
- This was changed to unsafe_load with the swap to YAMLSerializer. But
this method did not previously do an unsafe load and we shouldn't
provide that.
https://github.com/ruby/rubygems/commit/37f71c1eac - 03:52 AM Revision 1424201b (git): [ruby/rubygems] Add test for !binary tag in sequence item inline
- https://github.com/ruby/rubygems/commit/ba4a4b2b26
Co-authored-by: hsbt <12301+hsbt@users.noreply.github.com> -
- Decode base64-encoded values tagged with !binary in mapping keys,
mapping values (both inline and block scalar), and sequence items.
This fixes gem install failures for gems packaged with older RubyGems
that used !binary encoding in chec... -
- When a gem has native extensions and uses `require_relative` to load
a path without a corresponding .rb file in the gem, warn that this
will break in RubyGems 4.2, which will stop copying compiled
extensions into the gem's lib directory.... -
- Many gems created with the default template over the past 6 years use
`require_relative` to load compiled extensions, which breaks when
extensions are not copied into the gem's lib directory. Restore the
default to true for now to mainta... - 02:39 AM Revision f3083308 (git): Bump the github-actions group across 1 directory with 2 updates
- Bumps the github-actions group with 2 updates in the / directory: [ruby/setup-ruby](https://github.com/ruby/setup-ruby) and [advanced-security/filter-sarif](https://github.com/advanced-security/filter-sarif).
Updates `ruby/setup-ruby` ... -
- It appears that shims for individual commands are no longer being
created. -
- Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
-
- Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
-
- OpenBSD requires MAP_STACK for memory regions used as thread stacks.
However it seems to error with "Invalid argument" unless the permissions
include both PROT_READ | PROT_WRITE.
We should be able to satisft this by re-mmapping over our... -
12:38 AM Feature #21700: `IO::Buffer.map`: offset argument is "broken" and needs to be made more useful
- I've been using `IO::Buffer.map` on macOS without issues. It's still early days but I wonder why it was failing for you.
https://github.com/socketry/async-utilization/blob/main/lib/async/utilization/observer.rb
03/11/2026
-
11:49 PM Revision aa604d5a (git): [DOC] Fix indentation in docs for File.path
-
-
- https://github.com/ruby/prism/commit/2e58c52196
-
- https://github.com/ruby/prism/commit/3bdd79257b
-
04:02 PM Feature #21950: Add a built-in CPU-time profiler
- The opentelemetry community project includes people from a number of companies building profilers (datadog, google, elastic, polar signals) and it's proposing a new format that's inspired by pprof but tries to solve for some of its short...
-
- > conceptually it's like collecting data equivalent to caller_locations efficiently every X milliseconds.
> ...
Unless I've read this issue completely wrong, it is not about adding a sampling profiler, it is about adding a CPU time prof... -
- Eregon (Benoit Daloze) wrote in #note-6:
> > I am rather doubtful that CRuby/JRuby/TruffleRuby can share any profiler implementation at all.
> ...
"Profiler implementation" meant (1) the sampling backend and (2) what information would ... -
- osyoyu (Daisuke Aritomo) wrote in #note-5:
> I am rather doubtful that CRuby/JRuby/TruffleRuby can share any profiler implementation at all.
They can't share the implementation of gathering backtraces (that's implementation-specific)... -
- osyoyu (Daisuke Aritomo) wrote in #note-5:
> I am rather doubtful that CRuby/JRuby/TruffleRuby can share any profiler implementation at all.
I think the opposite, I think it's actually easy to implement such an API on TruffleRuby and... -
- Thank you for your comments and input!
Eregon (Benoit Daloze) wrote in #note-4:
> The API should not be under RubyVM, that's CRuby-specific
`RubyVM` being CRuby-specific is actually the exact reason I choosed the place to locate a... -
- I think this would be good, and I agree this would avoid several gems reimplementing this in brittle and not portable (e.g. only works on CRuby) ways.
See https://github.com/truffleruby/truffleruby/issues/2044#issuecomment-654848324 for... -
- > Any CPU timing in Ruby must also consider JIT improvements over time, and be able to decode things like inlined method calls so that the reported execution time is associated with the correct body of code.
Yes, ZJIT probabaly would ... - 01:24 PM Revision d72a0fed (git): Always take th->interrupt_lock in ubf_clear
- Patch 08372635f7 fixed a race condition on ubfs, but it's only valid if right after
a call to `ubf_clear`, we assume the ubf function cannot be in the middle of running.
This patch removes an optimization in `ubf_clear` that violates tha... -
- Thank you for the quick fix!
-
- Applied in changeset commit:git|679e4069217104f4fe1d887ec48aa8c3537245b2.
----------
Suppress bundled gems warning for subfeatures found outside stdlib [Bug #21828]
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> -
- A main difference between both reproductions are the Gemfile.lock, for the reproduction in the issue description:
```
GEM
remote: https://rubygems.org/
specs:
benchmark-ips (2.14.0)
PLATFORMS
ruby
x86_64-linux
DE... -
- Interestingly it seems fixed for the original reproduction in the issue description, but *not* for the new reproduction of running `benchmark-ips`'s tests.
-
- Also happens on `4.0-dev` from `ruby-build` (`ruby 4.0.1 (2026-03-10 revision 21f2bbf322) +PRISM [x86_64-linux]`).
-
- As I said:
Eregon (Benoit Daloze) wrote in #note-6:
> still a problem on master
I have tried with `ruby 4.1.0dev (2026-03-11T07:46:22Z master c08f977ac7) +PRISM [x86_64-linux]` specifically.
Please try the reproduction before closing. -
- The additional changes has been merged. I believe they fixed that.
* https://github.com/ruby/ruby/pull/15908
* https://github.com/ruby/ruby/pull/16154
Please look the merged commit at stable branch at first. -
- Re-opening since it's still a problem on master and on 4.0.1, so the fix doesn't seem to work.
Another reproduction from https://github.com/evanphx/benchmark-ips/pull/148 is:
```
git clone git@github.com:evanphx/benchmark-ips.git
cd ... -
- ## Summary
I investigate https://github.com/ruby/rubygems/issues/3799 for speed up Ruby's boot time. It depends on the environment, but when gems are included in `GEM_HOME` and user installation. I and Claude found that the most time-... -
- Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
-
- Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
-
09:51 AM Revision 76cca827 (git): Bump v3.4.9
- 07:46 AM Revision c08f977a (git): ZJIT: Add StringEqual HIR and support fold constants (#16106)
- Add a `StringEqual` Insn and use it to enable safe constant folding for string equality.
Constant folding uses `chase_insn` to normalize operands. We fold only when it is provably safe:
- both operands resolve to the same SSA value
- bo... -
- https://github.com/ruby/rubygems/commit/80e37bb0a6
-
- Unless of course fork isn't available.
Alternate: #7930, #7933
Fix: #7930, #7933
When bundler inline has to install gems, it loads more dependencies than when it
goes through the fast path of all gems being installed.
One of them is `... -
- https://github.com/ruby/rubygems/commit/4ce1510a60
-
- `long` and fixnum are not always sufficiently large for pointers.
-
-
- https://github.com/ruby/prism/commit/1d6a9669dc
-
-
- Move all the logic from prism.c into regexp.c. Now regexp.c does two passes. The first pass scans the raw source to track escape types, non-ASCII literals, and multibyte validity for encoding validation. The second pass scans the unescap...
-
- I think I've found a solution to this: we can return to the Ruby 3.4 `O(1)` removal, remove box/namespacing from it, and actually make it even simpler by skipping the CLASS -> ICLASS (and ICLASS -> ICLASS) relationship and directly assoc...
- 02:40 AM Revision 68bf517f (git): Bump taiki-e/install-action
- Bumps the github-actions group with 1 update in the / directory: [taiki-e/install-action](https://github.com/taiki-e/install-action).
Updates `taiki-e/install-action` from 2.68.25 to 2.68.26
- [Release notes](https://github.com/taiki-e... -
03/10/2026
-
08:27 PM Bug #21926: Thread#value on popen3 wait thread hangs in finalizer
- A note to whoever backports this (which could be me, I'm just waiting ~ 1 week and checking various CIs after the merge):
https://github.com/ruby/ruby/pull/16362 should be backported as well, they are really 1 fix in 2 separate commit... -
07:53 PM Revision 3fa74bff (git): ZJIT: Only emit IncrCounter instructions in stats mode (#16330)
- * ZJIT: Only emit IncrCounter instructions in stats mode
This cleans up a lot of HIR output when we're not explicitly measuring
counters. They don't generate code when not in stats mode so they are
just HIR clutter.
* ZJIT: Enable stat... -
- Also worth pointing out the ongoing efforts by the JIT team to eliminate artificial stack frames for leaf methods and some core method calls. As more artificial Ruby frames get elided, it will become harder to reconstruct the stack in a ...
-
- > JFR folks have decided to control sampling timing within the runtime to improve accuracy and stability
The other issue with real CPU timing is the heavy responsibility for accurately representing individual methods' CPU times withou... -
- Modern CRuby workloads can consume CPU concurrently across multiple native threads, especially with multiple Ractors and C exts which release the GVL. I'd like to propose the idea of integrating a built-in CPU-time profiler CRuby to enab...
-
- Applied in changeset commit:git|5b464d079afb4aad88ac9b9e21c6c4a9c3fb5c1d.
----------
[ruby/prism] Fix not binding power in endless methods
[Bug #21097]
Also, parse:
```
x = a rescue b in c
```
as:
```
(x = (a rescue b)) in c
```
a... -
- [Bug #21097]
Also, parse:
```
x = a rescue b in c
```
as:
```
(x = (a rescue b)) in c
```
and:
```
def f = a rescue b in c
```
as:
```
(def f = (a rescue b)) in c
```
https://github.com/ruby/prism/commit/7e797f59ae -
- https://github.com/ruby/prism/commit/979251ef06
-
- Windows does not have comm.exe, but it does have sort.exe. However,
the default Windows sort.exe is always case-insensitive, so its
outputs is not suitable for input to UNIX-like `comm` tool. -
- https://github.com/ruby/rubygems/commit/368fc29ce1
-
- https://github.com/ruby/rubygems/commit/2636b4d24a
-
- https://github.com/ruby/rubygems/commit/20153ebc78
-
- https://github.com/ruby/rubygems/commit/ba8e4ed9ed
-
- https://github.com/ruby/rubygems/commit/d4c5fc58e7
-
- https://github.com/ruby/rubygems/commit/b8f77e3860
-
- https://github.com/ruby/rubygems/commit/de53449860
-
- Cover native extensions and Windows-style paths
Verify files, extensions, and backslash fields are preserved by SafeYAML
https://github.com/ruby/rubygems/commit/5adce1d26d -
- https://github.com/ruby/rubygems/commit/e5bcde365f
-
- https://github.com/ruby/rubygems/commit/6b0aa2a2d3
-
- https://github.com/ruby/rubygems/commit/074e3f7f62
-
- https://github.com/ruby/rubygems/commit/cf566c3079
-
- https://github.com/ruby/rubygems/commit/cbe57bbbd7
-
- https://github.com/ruby/rubygems/commit/c29b3b2240
-
- https://github.com/ruby/rubygems/commit/940358f2c2
-
- https://github.com/ruby/rubygems/commit/c091444047
-
- https://github.com/ruby/rubygems/commit/5a422dafb1
- 07:29 AM Revision 3071dcf4 (git): Bump zizmorcore/zizmor-action
- Bumps the github-actions group with 1 update in the / directory: [zizmorcore/zizmor-action](https://github.com/zizmorcore/zizmor-action).
Updates `zizmorcore/zizmor-action` from 0.5.1 to 0.5.2
- [Release notes](https://github.com/zizmo... - 07:05 AM Revision 20c82234 (git): [DOC] Update bundled gems list at ae3ad5e13d8e4a2532c69f251b92a7
-
-
-
- It appears that shims for individual commands are no longer being
created. -
- https://github.com/ruby/rubygems/commit/c7bd0c4d5e
-
- https://github.com/ruby/rubygems/commit/75985f0c77
-
- https://github.com/ruby/rubygems/commit/733d9c9b0b
-
- (https://github.com/ruby/strscan/pull/196)
I noticed this while working on #195.
https://github.com/ruby/strscan/commit/3dc627db11 -
- This leads installation issue of Windows.
https://github.com/ruby/rubygems/commit/244b38acd7 -
- I personally think `eof?` being blocking is extremely confusing.
I use this shim: https://github.com/socketry/io-stream/blob/main/lib/io/stream/shim/readable.rb#L11 to avoid blocking.
I also mentioned the issue here: https://bugs.ruby-... -
- https://github.com/ruby/prism/commit/fd0a20f8d8
-
- * [Bug #21870] Regexp: Warnings when using slightly overlapping `\p{...}` classes (jneen)
* Warning spam on code that definitely isn't a mistake (`/[\p{Word}\p{S}]/` and other overlapping properties)
* Noted some possible ways forwar...