On Thursday, 4 March 2021 at 20:14:32 UTC, Curious George wrote:
> [...]

I'd like to apologize for my initial post as it wasn't sufficiently substantiated and thus resulted in an argument. Further, I am quite thankful to mipri for not only elaborating on my point, but expressing impressive meditative acumen. In the future, I hope to avoid this and actually defend my positions myself.

P.S. I'm mainly a D programming, but I've been looking into nim as of late, and I got a tad excited as one can see. As such, please think poorly of me for that post, but not nim's community.

On Friday, 5 March 2021 at 18:22:31 UTC, mipri wrote:
>   James Lu: Python has first-class C integration, and D has
>     first-class extern(C) support.
>
>   mipri: [These examples from Nim] are first class C
>    integration. What D has is very fine, but it's general seating.
>    [Python] was such a chore that http://swig.org/ happened to
>    pull the Python programmers off the wing

from ctypes import *
sox = cdll.LoadLibrary("libc.so.6")
sox.sox_init()

No headers required.

On 3/4/21 4:57 AM, Chris Piker wrote:
...
> code.  Others on these forums have mentioned that a good place to expand D's user base is the current crop of python programmers.  At least for folks like me who also maintain C libraries for performance, that sounds like a great focus area.
...
> Take care and I hope you keep up the good work, especially in the area of Linux IDE development.
> 

Chris, thanks for the note. We found the same: Our computational biology group are all python alumni, and the three of us on the software side have all switched to D. The transition from python was pretty seamless, cognitively, and I would say our overall productivity has gone UP.

Still have to use pandas for some dataframe type work, but otherwise, all code is written in D.

On Wednesday, 10 March 2021 at 03:06:40 UTC, James Blachly wrote:
> On 3/4/21 4:57 AM, Chris Piker wrote:
> ...
>> code.  Others on these forums have mentioned that a good place to expand D's user base is the current crop of python programmers.  At least for folks like me who also maintain C libraries for performance, that sounds like a great focus area.
> ...
>> Take care and I hope you keep up the good work, especially in the area of Linux IDE development.
>> 
>
> Chris, thanks for the note. We found the same: Our computational biology group are all python alumni, and the three of us on the software side have all switched to D. The transition from python was pretty seamless, cognitively, and I would say our overall productivity has gone UP.
>
> Still have to use pandas for some dataframe type work, but otherwise, all code is written in D.

Yeah, if more people just knew about D.

Me and another user converted some python code and got execution time down from hours (stopped the execution after x hours) to under 1 minute (I think it was about 20 seconds on my system iirc).

It was a very pleasing experience.

On Thursday, 4 March 2021 at 10:20:15 UTC, Iain Buclaw wrote:
> On Thursday, 4 March 2021 at 09:57:01 UTC, Chris Piker wrote:
>> The only downsides I've experienced with my current round of D programming is that:
>>
>> A) One of my programs compiles/runs fine under dmd, but the exact same code exhibits stack corruption under GDC-10.  (that was an interesting debugging session)  I haven't tried it under LDC yet.
>>
>
> Hi Chris,
>
> Thanks for sharing your experience.
>
> Would it be possible to raise a bug report with either a reduced test or steps to reproduce?
>
> https://gcc.gnu.org/bugzilla/enter_bug.cgi?product=gcc&component=d&version=10.0
>
> Regards,
> Iain.

Hi Iain

To get back to you on the stack corruption issue, I noticed some bugs in a dependency library I wrote in C.  I was using my old C library in new ways now that I have D bindings and ended up flushing some bugs out of the woodwork.  When I fixed the bugs, the stack corruption issue under GDC disappeared.  Not sure why they didn't appear in the DMD binaries, but who knows, vagaries of stack layout I guess.

Thanks for the quick reply to my initial message, sorry about the false alarm.

Also, appreciate your work on GDC!  I wish GDC was included in the package list for CentOS 8 (our production servers) but that's true of many packages that I end up building from source on CentOS.

Take Care,

On Wednesday, 10 March 2021 at 12:03:05 UTC, Imperatorn wrote:
> On Wednesday, 10 March 2021 at 03:06:40 UTC, James Blachly
>
> Yeah, if more people just knew about D.

Without a big corporate backer adoption is slow, but I hope the D community just keeps plugging along anyway.

> Me and another user converted some python code and got execution time down from hours (stopped the execution after x hours) to under 1 minute (I think it was about 20 seconds on my system iirc).
>
> It was a very pleasing experience.

Similar story here.  I provide a streaming service for a few space particles and fields research groups.  Some of the radio spectrum generators were taking more than 35 minutes to provide a day's worth of data, thus causing client programs to time out.  It was frustrating because the FFTs ran in fortran code after all, so python wasn't even doing most of the work.

After converting the streamer programs to D, CPU utilization is less than 30% of a single core (even on debug builds on DMD) while maxing out the RAID-5 array.  Response time is now less than 18 seconds, which is still long but it's now a hardware limitation, not a software problem.

To me D feels like an upgrade, but not a radical break.  Both D and python have modules.  Associative arrays swap in for dictionaries.  Range operations feel like list comprehensions, and I'm sure there are more similarities I've yet to encounter.

The learning curve is a bit steep at times, though Ali's book has been very helpful in that regard and the gdc+gdb+ddd stack has provided a reasonable debugging experience so far.

Happy Programming,

On Friday, 12 March 2021 at 07:50:52 UTC, Chris Piker wrote:
> On Wednesday, 10 March 2021 at 12:03:05 UTC, Imperatorn wrote:
> it's now a hardware limitation, not a software problem

I havent seen the data but most measurement data is in narrow range so delta compression could save a lot of space. Just an idea.

On Wednesday, 10 March 2021 at 12:03:05 UTC, Imperatorn wrote:

> Yeah, if more people just knew about D.
>
> Me and another user converted some python code and got execution time down from hours (stopped the execution after x hours) to under 1 minute (I think it was about 20 seconds on my system iirc).
>
> It was a very pleasing experience.

It would be great to write a short article on the transition from Python to D, or even a details manual