• Walter Bright (20/20) Jun 25 2016 Andrei identified a key blocker for D adoption is the incomplete impleme...
• Walter Bright (5/5) Jun 25 2016 Debugging tip:
• Walter Bright (3/4) Jun 25 2016 A unittest that is deliberately unsafe should be annotated with @system....
• Tourist (3/8) Jun 25 2016 So why not declare @safe module-wise, and only add @system
• Jonathan M Davis via Digitalmars-d (10/19) Jun 25 2016 In general, IMHO, mass applying attributes is a bad idea, because it mak...
• Atila Neves (3/8) Jun 29 2016 What about existing tests that are marked @trusted?
• Lodovico Giaretta (7/16) Jun 29 2016 Maybe I'm missing something, but...
• Atila Neves (5/23) Jun 29 2016 I don't think there's any point either, I just thought I'd ask
• Walter Bright (2/10) Jun 29 2016 Marking a unittest as @trusted seems pointless.
• Steven Schveighoffer (5/17) Jun 30 2016 I would further say -- if you see a @trusted unit test, whoever wrote it...
• Walter Bright (3/6) Jun 30 2016 I agree. Using @trusted properly turns out to have some surprising subtl...
• Andre Pany (9/16) Jun 26 2016 Could we have a wiki page for this call to action? It would be
• Robert burner Schadek (4/4) Jun 26 2016 It would be awesome if you would create that process model in the
• Walter Bright (2/5) Jun 26 2016 It's a wiki, feel free to add it.
• Robert burner Schadek (11/12) Jun 27 2016 I have to say that reply really makes me angry. I created that
• Walter Bright (4/15) Jun 27 2016 Sorry to have offended you, I worded things badly. Thank you for making ...
• Robert burner Schadek (6/10) Jun 28 2016 Don't sweat it. I just replied in the heat of the moment and
• Andrei Alexandrescu (3/13) Jun 28 2016 That's a great rule. I should add that the official "things to do" list
• Bennet Leff (5/9) Jun 30 2016 Could you elaborate on list option 9 "create a module that
• Walter Bright (3/11) Jun 30 2016 I don't know. John Colvin is working on this, he could give a far better...
• Seb (4/20) Jun 30 2016 Some pointers:
• Bennet Leff (3/19) Jul 01 2016 Thanks I'll take a further look although it doesn't look like his
• Smoke (4/10) Jun 26 2016 How hard would it be to add a compiler switch, similar to -vgs
• Atila Neves (12/34) Jul 14 2016 It's possible that someone introduced new unittests that aren't
• Walter Bright (2/10) Jul 15 2016 Awesome, thank you!

Walter Bright <newshound2 digitalmars.com> writes:

Andrei identified a key blocker for D adoption is the incomplete implementation 
of  safe. I'm working on the compiler end. But Phobos has a lot of code that is 
pointlessly not  safe, making it frustrating to write  safe code that calls 
Phobos. Some are listed in Bugzilla, most are not.

So here's what to do:

1. Pick a module, any module, let's say 'std.foo'.

2. Compile it:

     dmd -unittest -main std/foo

and run:

     foo

to verify that the unittests work.

3. Take a look at foo.d and look for a unittest that is not marked  safe.

4. Add  safe to the unittest

5. Do Step 2 on it.

6. If it works, submit a PR with the annotation (well, try to collect a few of 
these in one module)

7. If it doesn't work, find out where the unsafe code is and fix it, and submit 
a PR.


I've submitted maybe a dozen PRs against Phobos over the last day from doing 
this. It's only a start.

Walter Bright <newshound2 digitalmars.com> writes:

Debugging tip:

Templates should almost never be marked  safe or  trusted, as the safety should 
be inferred. However, to isolate down where the unsafe code actually is, add 
 safe temporarily to the template declaration. Then, the compiler will tell you 
which line in the template is unsafe.

Walter Bright <newshound2 digitalmars.com> writes:

On 6/25/2016 3:44 PM, Walter Bright wrote:
 4. Add  safe to the unittest

A unittest that is deliberately unsafe should be annotated with  system.
Meaning 
that any un-annotated unittest needs corrective action one way or the other.

Tourist <gravatar gravatar.com> writes:

On Saturday, 25 June 2016 at 22:56:12 UTC, Walter Bright wrote:
 On 6/25/2016 3:44 PM, Walter Bright wrote:
 4. Add  safe to the unittest

 A unittest that is deliberately unsafe should be annotated with 
  system. Meaning that any un-annotated unittest needs 
 corrective action one way or the other.

So why not declare  safe module-wise, and only add  system 
annotations where needed?

Jonathan M Davis via Digitalmars-d <digitalmars-d puremagic.com> writes:

On Sunday, June 26, 2016 06:29:05 Tourist via Digitalmars-d wrote:
 On Saturday, 25 June 2016 at 22:56:12 UTC, Walter Bright wrote:
 On 6/25/2016 3:44 PM, Walter Bright wrote:
 4. Add  safe to the unittest

 A unittest that is deliberately unsafe should be annotated with
  system. Meaning that any un-annotated unittest needs
 corrective action one way or the other.

 So why not declare  safe module-wise, and only add  system
 annotations where needed?

In general, IMHO, mass applying attributes is a bad idea, because it makes
it harder to know which attributes apply where, which is a maintenance
problem. However, even worse than that, while there it _is_ possible to
apply  safe to the module as a whole and then selectively apply  trusted and
 system explicitly, there is no way to turn off all of the  safety
attributes so that attribute inference can be used. And other attributes
can't be undone at all. So, if you have templates in a module, mass applying
attributes is a disaster.

- Jonathan M Davis

Atila Neves <atila.neves gmail.com> writes:

On Saturday, 25 June 2016 at 22:56:12 UTC, Walter Bright wrote:
 On 6/25/2016 3:44 PM, Walter Bright wrote:
 4. Add  safe to the unittest

 A unittest that is deliberately unsafe should be annotated with 
  system. Meaning that any un-annotated unittest needs 
 corrective action one way or the other.

What about existing tests that are marked  trusted?

Atila

Lodovico Giaretta <lodovico giaretart.net> writes:

On Wednesday, 29 June 2016 at 17:25:31 UTC, Atila Neves wrote:
 On Saturday, 25 June 2016 at 22:56:12 UTC, Walter Bright wrote:
 On 6/25/2016 3:44 PM, Walter Bright wrote:
 4. Add  safe to the unittest

 A unittest that is deliberately unsafe should be annotated 
 with  system. Meaning that any un-annotated unittest needs 
 corrective action one way or the other.

 What about existing tests that are marked  trusted?

 Atila

Maybe I'm missing something, but...
What's the purpose of a  trusted unittest? If it's safe, mark it 
 safe. If it's not, you gain nothing from marking it  trusted, 
given that it will not be called by  safe code (it will not be 
called in general).  trusting a top-level function is useless 
IMHO...

Atila Neves <atila.neves gmail.com> writes:

On Wednesday, 29 June 2016 at 17:33:18 UTC, Lodovico Giaretta 
wrote:
 On Wednesday, 29 June 2016 at 17:25:31 UTC, Atila Neves wrote:
 On Saturday, 25 June 2016 at 22:56:12 UTC, Walter Bright wrote:
 On 6/25/2016 3:44 PM, Walter Bright wrote:
 4. Add  safe to the unittest

 A unittest that is deliberately unsafe should be annotated 
 with  system. Meaning that any un-annotated unittest needs 
 corrective action one way or the other.

 What about existing tests that are marked  trusted?

 Atila

 Maybe I'm missing something, but...
 What's the purpose of a  trusted unittest? If it's safe, mark 
 it  safe. If it's not, you gain nothing from marking it 
  trusted, given that it will not be called by  safe code (it 
 will not be called in general).  trusting a top-level function 
 is useless IMHO...

I don't think there's any point either, I just thought I'd ask 
before changing them to  system.

Atila

Walter Bright <newshound2 digitalmars.com> writes:

On 6/29/2016 10:25 AM, Atila Neves wrote:
 On Saturday, 25 June 2016 at 22:56:12 UTC, Walter Bright wrote:
 On 6/25/2016 3:44 PM, Walter Bright wrote:
 4. Add  safe to the unittest

 A unittest that is deliberately unsafe should be annotated with  system.
 Meaning that any un-annotated unittest needs corrective action one way or the
 other.

 What about existing tests that are marked  trusted?

Marking a unittest as  trusted seems pointless.

Steven Schveighoffer <schveiguy yahoo.com> writes:

On 6/29/16 4:57 PM, Walter Bright wrote:
 On 6/29/2016 10:25 AM, Atila Neves wrote:
 On Saturday, 25 June 2016 at 22:56:12 UTC, Walter Bright wrote:
 On 6/25/2016 3:44 PM, Walter Bright wrote:
 4. Add  safe to the unittest

 A unittest that is deliberately unsafe should be annotated with  system.
 Meaning that any un-annotated unittest needs corrective action one
 way or the
 other.

 What about existing tests that are marked  trusted?

 Marking a unittest as  trusted seems pointless.

I would further say -- if you see a  trusted unit test, whoever wrote it 
doesn't understand the point of  trusted. Be sure and check the code it 
is testing too, and file bugs if appropriate.

-Steve

Walter Bright <newshound2 digitalmars.com> writes:

On 6/30/2016 4:09 AM, Steven Schveighoffer wrote:
 I would further say -- if you see a  trusted unit test, whoever wrote it
doesn't
 understand the point of  trusted. Be sure and check the code it is testing too,
 and file bugs if appropriate.

I agree. Using  trusted properly turns out to have some surprising subtleties, 
but this one is pretty much a no-brainer.

Andre Pany <andre s-e-a-p.de> writes:

On Saturday, 25 June 2016 at 22:44:37 UTC, Walter Bright wrote:
 Andrei identified a key blocker for D adoption is the 
 incomplete implementation of  safe. I'm working on the compiler 
 end. But Phobos has a lot of code that is pointlessly not 
  safe, making it frustrating to write  safe code that calls 
 Phobos. Some are listed in Bugzilla, most are not.

 So here's what to do:

 1....

Could we have a wiki page for this call to action? It would be 
much easier if there is a list of the modules and their current 
state like: completed or who is working on this module.

In general the call to actions topics disappears too fast in a 
forum. Maybe it worths to add them to the contributions group in 
the wiki?

Kind regards
Andre

Robert burner Schadek <rburners gmail.com> writes:

It would be awesome if you would create that process model in the 
wiki and at it to your action list 
http://wiki.dlang.org/Walter_Andrei_Action_List#Walter_and_Andrei.27s_Action_List

So it does not get lost and people can find it.

Walter Bright <newshound2 digitalmars.com> writes:

On 6/26/2016 6:13 AM, Robert burner Schadek wrote:
 It would be awesome if you would create that process model in the wiki and at
it
 to your action list
 http://wiki.dlang.org/Walter_Andrei_Action_List#Walter_and_Andrei.27s_Action_List


It's a wiki, feel free to add it.

Robert burner Schadek <rburners gmail.com> writes:

On Sunday, 26 June 2016 at 22:38:54 UTC, Walter Bright wrote:
 It's a wiki, feel free to add it.

I have to say that reply really makes me angry. I created that 
list so Andrei and you have an easy to find spot where you can 
write down tasks so people can work on them. You did not disagree 
with the list at the time, and didn't disagree so far. I really 
think it is clear that these are two list that you two are 
maintaining, how else should anybody know what work is 
pre-approved and where D is going.

I think I'm not asking too much here, if I ask you to decide if 
you want to maintain that list or if you want to delete it. The 
current limbo state is just making things worse.

Walter Bright <newshound2 digitalmars.com> writes:

On 6/27/2016 8:14 AM, Robert burner Schadek wrote:
 On Sunday, 26 June 2016 at 22:38:54 UTC, Walter Bright wrote:
 It's a wiki, feel free to add it.

 I have to say that reply really makes me angry. I created that list so Andrei
 and you have an easy to find spot where you can write down tasks so people can
 work on them. You did not disagree with the list at the time, and didn't
 disagree so far. I really think it is clear that these are two list that you
two
 are maintaining, how else should anybody know what work is pre-approved and
 where D is going.

 I think I'm not asking too much here, if I ask you to decide if you want to
 maintain that list or if you want to delete it. The current limbo state is just
 making things worse.

Sorry to have offended you, I worded things badly. Thank you for making the 
list. It's just that I'm feeling a bit overwhelmed at the moment with trying to 
get things done and being asked to do more every day, and I'd like to delegate.

Robert burner Schadek <rburners gmail.com> writes:

On Monday, 27 June 2016 at 19:33:45 UTC, Walter Bright wrote:
 Sorry to have offended you, I worded things badly. Thank you 
 for making the list. It's just that I'm feeling a bit 
 overwhelmed at the moment with trying to get things done and 
 being asked to do more every day, and I'd like to delegate.

Don't sweat it. I just replied in the heat of the moment and 
thereby violated my personal 24 hour no reply rule.

Back to topic, I think this list will save you and Andrei work in 
the long run. I mean if the list is up to date, you do not have 
to repeat yourself. Well expect "read the list!" ;-)

Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> writes:

On 06/28/2016 05:34 PM, Robert burner Schadek wrote:
 On Monday, 27 June 2016 at 19:33:45 UTC, Walter Bright wrote:
 Sorry to have offended you, I worded things badly. Thank you for
 making the list. It's just that I'm feeling a bit overwhelmed at the
 moment with trying to get things done and being asked to do more every
 day, and I'd like to delegate.

 Don't sweat it. I just replied in the heat of the moment and thereby
 violated my personal 24 hour no reply rule.

 Back to topic, I think this list will save you and Andrei work in the
 long run. I mean if the list is up to date, you do not have to repeat
 yourself. Well expect "read the list!" ;-)

That's a great rule. I should add that the official "things to do" list 
is the vision document. We don't need a separate list. -- Andrei

Bennet Leff <bennetleff gmail.com> writes:

On Sunday, 26 June 2016 at 13:13:01 UTC, Robert burner Schadek 
wrote:
 It would be awesome if you would create that process model in 
 the wiki and at it to your action list 
 http://wiki.dlang.org/Walter_Andrei_Action_List#Walter_and_Andrei.27s_Action_List

 So it does not get lost and people can find it.

Could you elaborate on list option 9 "create a module that 
enables code to be run on GPU." Wouldn't the Derelict OpenCL 
bindings satisfy that need?

Walter Bright <newshound2 digitalmars.com> writes:

On 6/30/2016 11:54 AM, Bennet Leff wrote:
 On Sunday, 26 June 2016 at 13:13:01 UTC, Robert burner Schadek wrote:
 It would be awesome if you would create that process model in the wiki and at
 it to your action list
 http://wiki.dlang.org/Walter_Andrei_Action_List#Walter_and_Andrei.27s_Action_List

 So it does not get lost and people can find it.

 Could you elaborate on list option 9 "create a module that enables code to be
 run on GPU." Wouldn't the Derelict OpenCL bindings satisfy that need?

I don't know. John Colvin is working on this, he could give a far better 
appraisal of the state.

Seb <seb wilzba.ch> writes:

On Thursday, 30 June 2016 at 21:31:25 UTC, Walter Bright wrote:
 On 6/30/2016 11:54 AM, Bennet Leff wrote:
 On Sunday, 26 June 2016 at 13:13:01 UTC, Robert burner Schadek 
 wrote:
 It would be awesome if you would create that process model in 
 the wiki and at
 it to your action list
 http://wiki.dlang.org/Walter_Andrei_Action_List#Walter_and_Andrei.27s_Action_List

 So it does not get lost and people can find it.

 Could you elaborate on list option 9 "create a module that 
 enables code to be
 run on GPU." Wouldn't the Derelict OpenCL bindings satisfy 
 that need?

 I don't know. John Colvin is working on this, he could give a 
 far better appraisal of the state.

Some pointers:

https://github.com/John-Colvin/clWrap
http://dconf.org/2016/talks/colvin.html

Bennet Leff <bennetleff gmail.com> writes:

On Thursday, 30 June 2016 at 21:55:33 UTC, Seb wrote:
 On Thursday, 30 June 2016 at 21:31:25 UTC, Walter Bright wrote:
 On 6/30/2016 11:54 AM, Bennet Leff wrote:
 On Sunday, 26 June 2016 at 13:13:01 UTC, Robert burner 
 Schadek wrote:
 [...]

 Could you elaborate on list option 9 "create a module that 
 enables code to be
 run on GPU." Wouldn't the Derelict OpenCL bindings satisfy 
 that need?

 I don't know. John Colvin is working on this, he could give a 
 far better appraisal of the state.

 Some pointers:

 https://github.com/John-Colvin/clWrap
 http://dconf.org/2016/talks/colvin.html

Thanks I'll take a further look although it doesn't look like his 
project is ready for more contributors to add to it yet.

"Smoke" Adams <SA2432 gmail.com> writes:

On Saturday, 25 June 2016 at 22:44:37 UTC, Walter Bright wrote:
 Andrei identified a key blocker for D adoption is the 
 incomplete implementation of  safe. I'm working on the compiler 
 end. But Phobos has a lot of code that is pointlessly not 
  safe, making it frustrating to write  safe code that calls 
 Phobos. Some are listed in Bugzilla, most are not.

 [...]


How hard would it be to add a compiler switch, similar to -vgs 
that shows every non-safe function? This would not just help out 
with phobos.

Atila Neves <atila.neves gmail.com> writes:

On Saturday, 25 June 2016 at 22:44:37 UTC, Walter Bright wrote:
 Andrei identified a key blocker for D adoption is the 
 incomplete implementation of  safe. I'm working on the compiler 
 end. But Phobos has a lot of code that is pointlessly not 
  safe, making it frustrating to write  safe code that calls 
 Phobos. Some are listed in Bugzilla, most are not.

 So here's what to do:

 1. Pick a module, any module, let's say 'std.foo'.

 2. Compile it:

     dmd -unittest -main std/foo

 and run:

     foo

 to verify that the unittests work.

 3. Take a look at foo.d and look for a unittest that is not 
 marked  safe.

 4. Add  safe to the unittest

 5. Do Step 2 on it.

 6. If it works, submit a PR with the annotation (well, try to 
 collect a few of these in one module)

 7. If it doesn't work, find out where the unsafe code is and 
 fix it, and submit a PR.


 I've submitted maybe a dozen PRs against Phobos over the last 
 day from doing this. It's only a start.

It's possible that someone introduced new unittests that aren't 
explicitly  safe or  system, but with the exception of std.stream 
(because of imminent deprecation) all phobos unittests are now 
explicity tagged. So now, if you want to help make Phobos  safe, 
the steps are:

1. Pick a module, any module
2. Search for a  system unittest
3. Change  system to  safe
4. Figure out why it doesn't compile that way and either edit 
code or file bugs


Atila

Walter Bright <newshound2 digitalmars.com> writes:

On 7/14/2016 8:57 AM, Atila Neves wrote:
 It's possible that someone introduced new unittests that aren't explicitly
 safe
 or  system, but with the exception of std.stream (because of imminent
 deprecation) all phobos unittests are now explicity tagged. So now, if you want
 to help make Phobos  safe, the steps are:

 1. Pick a module, any module
 2. Search for a  system unittest
 3. Change  system to  safe
 4. Figure out why it doesn't compile that way and either edit code or file bugs

Awesome, thank you!