www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.announce - Beta 2.073.0-b1

reply Martin Nowak <code+news.digitalmars dawg.eu> writes:
First beta for the 2.073.0 release.

This release comes with a few phobos additions, a new -mcpu=avx switch,
an experimental safety checks (-transition=safe/-dip1000), and several
bugfixes.

http://dlang.org/download.html#dmd_beta
http://dlang.org/changelog/2.073.0.html

Please report any bugs at https://issues.dlang.org

-Martin
Jan 06 2017
next sibling parent Jon Degenhardt <noreply noreply.com> writes:
On Saturday, 7 January 2017 at 05:02:13 UTC, Martin Nowak wrote:

 First beta for the 2.073.0 release.

 This release comes with a few phobos additions, a new -mcpu=avx 
 switch, an experimental safety checks 
 (-transition=safe/-dip1000), and several bugfixes.

 http://dlang.org/download.html#dmd_beta 
 http://dlang.org/changelog/2.073.0.html

 Please report any bugs at https://issues.dlang.org

 -Martin


The change log should probably include the topN rewrite. PR 4815, 
several issue reports.

--Jon
Jan 06 2017
prev sibling next sibling parent Basile B. <b2.temp gmx.com> writes:
On Saturday, 7 January 2017 at 05:02:13 UTC, Martin Nowak wrote:

 First beta for the 2.073.0 release.

 This release comes with a few phobos additions, a new -mcpu=avx 
 switch, an experimental safety checks 
 (-transition=safe/-dip1000), and several bugfixes.

 http://dlang.org/download.html#dmd_beta 
 http://dlang.org/changelog/2.073.0.html

 Please report any bugs at https://issues.dlang.org

 -Martin


I don't know if any DMD developers has noticed this regression:

https://issues.dlang.org/show_bug.cgi?id=17059

2.073.0-b1 is still affected.
I ask because it was reported the during the DFeed outage last 
week.
Jan 06 2017
prev sibling next sibling parent reply Jacob Carlborg <doob me.com> writes:
On 2017-01-07 06:02, Martin Nowak wrote:


 an experimental safety checks (-transition=safe/-dip1000)


Is the flag to enable the new checks or revert to the previous behavior?

-- 
/Jacob Carlborg
Jan 07 2017
parent reply Walter Bright <newshound2 digitalmars.com> writes:
On 1/7/2017 3:54 AM, Jacob Carlborg wrote:

 On 2017-01-07 06:02, Martin Nowak wrote:

 an experimental safety checks (-transition=safe/-dip1000)


 Is the flag to enable the new checks or revert to the previous behavior?


enable
Jan 14 2017
parent Timothee Cour via Digitalmars-d-announce writes:
 * -transition=safe/-dip1000
 => not mentioned in http://dlang.org/changelog/2.073.0.html



 That's deliberate, as you can't even use it with writeln yet.


https://issues.dlang.org/show_bug.cgi?id=17090 `dmd -transition=?` needs
quoting => make it `-transition=help`

On Sat, Jan 14, 2017 at 3:46 AM, Walter Bright via Digitalmars-d-announce <
digitalmars-d-announce puremagic.com> wrote:


 On 1/7/2017 3:54 AM, Jacob Carlborg wrote:


 On 2017-01-07 06:02, Martin Nowak wrote:


 an experimental safety checks (-transition=safe/-dip1000)


 Is the flag to enable the new checks or revert to the previous behavior?


 enable
Jan 14 2017
prev sibling next sibling parent reply biozic <dransic gmail.com> writes:
On Saturday, 7 January 2017 at 05:02:13 UTC, Martin Nowak wrote:

 First beta for the 2.073.0 release.

 This release comes with a few phobos additions, a new -mcpu=avx 
 switch, an experimental safety checks 
 (-transition=safe/-dip1000), and several bugfixes.

 http://dlang.org/download.html#dmd_beta 
 http://dlang.org/changelog/2.073.0.html

 Please report any bugs at https://issues.dlang.org

 -Martin


Thanks for the good work!

Are phobos unittests not passing when compiling with -dip1000 
considered bugs or is it this experimental?

BTW, there is no '-transition=safe' switch.

--
biozic
Jan 07 2017
next sibling parent reply Martin Nowak <code+news.digitalmars dawg.eu> writes:
On 01/07/2017 03:44 PM, biozic wrote:

 On Saturday, 7 January 2017 at 05:02:13 UTC, Martin Nowak wrote:

 First beta for the 2.073.0 release.

 This release comes with a few phobos additions, a new -mcpu=avx
 switch, an experimental safety checks (-transition=safe/-dip1000), and
 several bugfixes.

 http://dlang.org/download.html#dmd_beta
 http://dlang.org/changelog/2.073.0.html

 Please report any bugs at https://issues.dlang.org

 -Martin


 
 Thanks for the good work!
 
 Are phobos unittests not passing when compiling with -dip1000 considered
 bugs or is it this experimental?


Most of phobos is currently not usable with DIP1000, partly because
safety checks are hard errors (instead of deprecations) and partly
because of implementation bugs.
You can try it experimentally with isolated code snippets.
Don't expect much documentation or help at this stage, the feature will
properly released when it's finished.

Any way to escape pointers in  safe code with -dip1000 enabled are
considered bugs and should be reported under issues.dlang.org using the
`safe` keyword and `[scope]` in the title.

https://issues.dlang.org/buglist.cgi?keywords=safe&keywords_type=allwords&query_format=advanced&resolution=---&short_desc=%5Bscope%5D&short_desc_type=substring

-Martin
Jan 07 2017
parent reply biozic <dransic gmail.com> writes:
On Saturday, 7 January 2017 at 15:59:47 UTC, Martin Nowak wrote:

 On 01/07/2017 03:44 PM, biozic wrote:

 Are phobos unittests not passing when compiling with -dip1000 
 considered bugs or is it this experimental?


 Most of phobos is currently not usable with DIP1000, partly 
 because
 safety checks are hard errors (instead of deprecations) and 
 partly
 because of implementation bugs.
 You can try it experimentally with isolated code snippets.
 Don't expect much documentation or help at this stage, the 
 feature will
 properly released when it's finished.

 Any way to escape pointers in  safe code with -dip1000 enabled 
 are considered bugs and should be reported under 
 issues.dlang.org using the `safe` keyword and `[scope]` in the 
 title.

 https://issues.dlang.org/buglist.cgi?keywords=safe&keywords_type=allwords&query_format=advanced&resolution=---&short_desc=%5Bscope%5D&short_desc_type=substring

 -Martin


Thanks for the precisions!
Jan 07 2017
parent reply Timothee Cour via Digitalmars-d-announce writes:
* -transition=safe/-dip1000
=> not mentioned in http://dlang.org/changelog/2.073.0.html

* std.experimental.ndslice has been deprecated. The synchronization between
Phobos and Mir turned out to be a lot of work with litte gain

=> That's a good move. Will make development faster indeed by not being
tied to dmd releases. The downside of `batteries included` is is hampers
development speed, deprecation speed, etc. Another example: std.json sucks
compared to http://vibed.org/api/vibe.data.json/



On Sat, Jan 7, 2017 at 8:49 AM, biozic via Digitalmars-d-announce <
digitalmars-d-announce puremagic.com> wrote:


 On Saturday, 7 January 2017 at 15:59:47 UTC, Martin Nowak wrote:


 On 01/07/2017 03:44 PM, biozic wrote:


 Are phobos unittests not passing when compiling with -dip1000 considered
 bugs or is it this experimental?


 Most of phobos is currently not usable with DIP1000, partly because
 safety checks are hard errors (instead of deprecations) and partly
 because of implementation bugs.
 You can try it experimentally with isolated code snippets.
 Don't expect much documentation or help at this stage, the feature will
 properly released when it's finished.

 Any way to escape pointers in  safe code with -dip1000 enabled are
 considered bugs and should be reported under issues.dlang.org using the
 `safe` keyword and `[scope]` in the title.

 https://issues.dlang.org/buglist.cgi?keywords=safe&keywords_
 type=allwords&query_format=advanced&resolution=---&short_
 desc=%5Bscope%5D&short_desc_type=substring

 -Martin


 Thanks for the precisions!
Jan 07 2017
parent Martin Nowak <code dawg.eu> writes:
On Saturday, 7 January 2017 at 21:39:18 UTC, Timothee Cour wrote:

 * -transition=safe/-dip1000
 => not mentioned in http://dlang.org/changelog/2.073.0.html


That's deliberate, as you can't even use it with writeln yet.


 * std.experimental.ndslice has been deprecated. The 
 synchronization between Phobos and Mir turned out to be a lot 
 of work with litte gain

 => That's a good move. Will make development faster indeed by 
 not being tied to dmd releases. The downside of `batteries 
 included` is is hampers development speed, deprecation speed, 
 etc. Another example: std.json sucks compared to 
 http://vibed.org/api/vibe.data.json/


Which has a much more powerful successor 
http://code.dlang.org/packages/std_data_json.
IMO, packages are great in std, once they're done and won't 
change much.
Jan 11 2017
prev sibling parent reply Walter Bright <newshound2 digitalmars.com> writes:
On 1/7/2017 6:44 AM, biozic wrote:

 Are phobos unittests not passing when compiling with -dip1000 considered bugs
or
 is it this experimental?


druntime is now being compiled with -dip1000.

   https://github.com/dlang/druntime/pull/1733

Phobos is next!
Jan 14 2017
parent =?UTF-8?B?Tm9yZGzDtnc=?= <per.nordlow gmail.com> writes:
On Sunday, 15 January 2017 at 06:07:23 UTC, Walter Bright wrote:

 Phobos is next!


I can't wait :)
Jan 15 2017
prev sibling next sibling parent reply bitwise <bitwise.pvt gmail.com> writes:
On Saturday, 7 January 2017 at 05:02:13 UTC, Martin Nowak wrote:

 First beta for the 2.073.0 release.

 This release comes with a few phobos additions, a new -mcpu=avx 
 switch, an experimental safety checks 
 (-transition=safe/-dip1000), and several bugfixes.

 http://dlang.org/download.html#dmd_beta 
 http://dlang.org/changelog/2.073.0.html

 Please report any bugs at https://issues.dlang.org

 -Martin


Is it possible to get an ETA on protection work on __traits?

Thanks
Jan 13 2017
parent reply Martin Nowak <code+news.digitalmars dawg.eu> writes:
On 01/14/2017 03:14 AM, bitwise wrote:

 Is it possible to get an ETA on protection work on __traits?


As any responsible programmer, I won't give out dates ;).
But as was mentioned this is mostly just removing existing access
checks, the visibility checks already treat __traits specially.
This can be done once the visibility changes are fully deprecated and
become errors. So far adoption was fairly slow though.

Changing things specifically for __traits(getMember, exp, "field") which
is currently lowered to exp.field would be too much effort.

-Martin
Jan 14 2017
parent reply Jacob Carlborg <doob me.com> writes:
On 2017-01-15 01:52, Martin Nowak wrote:


 Changing things specifically for __traits(getMember, exp, "field") which
 is currently lowered to exp.field would be too much effort.


What about lowering to "exp.tupleof[index]", which already bypass 
protection?

-- 
/Jacob Carlborg
Jan 15 2017
parent Martin Nowak <code+news.digitalmars dawg.eu> writes:
Sorry, this is OT in the beta thread.

On 01/15/2017 03:31 PM, Jacob Carlborg wrote:

 What about lowering to "exp.tupleof[index]", which already bypass
 protection?


We also have lots of other stuff to do, and no there is no 1-to-1
mapping between exp.name and exp.tupleof[index].
Jan 15 2017
prev sibling parent reply Basile B. <b2.temp gmx.com> writes:
On Saturday, 7 January 2017 at 05:02:13 UTC, Martin Nowak wrote:

 First beta for the 2.073.0 release.

 This release comes with a few phobos additions, a new -mcpu=avx 
 switch, an experimental safety checks 
 (-transition=safe/-dip1000), and several bugfixes.

 http://dlang.org/download.html#dmd_beta 
 http://dlang.org/changelog/2.073.0.html

 Please report any bugs at https://issues.dlang.org

 -Martin


DCD master doesn't build with this version. Try to build using 
make (and not DUB).
You should get, at the end:

[...]
bin/dcd-server.o: dans la fonction « _Dmain »:
msgpack-d/src/msgpack/value.d:(.text._Dmain+0xa38): référence 
indéfinie vers « 
_D4core4time12TickDuration25__T10opOpAssignVAyaa1_2bZ10opOpAssignMFNaNbNcNiNfS4core4time12TickDurationZS4cor
4time12TickDuration »
collect2: error: ld returned 1 exit status
Error: linker exited with status 1
makefile:124: recipe for target 'dmdserver' failed
make: *** [dmdserver] Error 1

I don't know what to think. With 2.072.2 no problem.
This doesn't look like a standard regression since it happens 
during linking.

Please someone test and confirm.
Jan 13 2017
parent reply Basile B. <b2.temp gmx.com> writes:
On Saturday, 14 January 2017 at 07:51:13 UTC, Basile B. wrote:

 On Saturday, 7 January 2017 at 05:02:13 UTC, Martin Nowak wrote:

 First beta for the 2.073.0 release.


 _D4core4time12TickDuration25__T10opOpAssignVAyaa1_2bZ10opOpAssignMFNaNbNcNiNfS4core4time12TickDurationZS4cor
4time12TickDuration »
 collect2: error: ld returned 1 exit status
 Error: linker exited with status 1
 makefile:124: recipe for target 'dmdserver' failed
 make: *** [dmdserver] Error 1

 I don't know what to think. With 2.072.2 no problem.
 This doesn't look like a standard regression since it happens 
 during linking.

 Please someone test and confirm.


Forget this, I've realized a bit late that it's already reported:
https://issues.dlang.org/show_bug.cgi?id=17072
Jan 14 2017
parent Martin Nowak <code+news.digitalmars dawg.eu> writes:
On 01/14/2017 09:10 AM, Basile B. wrote:

 Forget this, I've realized a bit late that it's already reported:
 https://issues.dlang.org/show_bug.cgi?id=17072


The cause is that we build druntime on Windows with -dip25.
https://issues.dlang.org/show_bug.cgi?id=17072#c1
Jan 15 2017