Send me your list of D gripes and wishes (page 6)

On Sunday, 25 December 2022 at 18:16:59 UTC, Richard (Rikki) Andrew Cattermole wrote: > We can't add things to Phobos because of leadership. In the last 10 years we have only added allocators and loggers to experimental, and only one of them was complete enough to come out of experimental. It has nothing at all to do with some flag why we can't put things in. It has always been a nightmare. I guess that's true if you're adding an entire module. When adding an individual function, or a new overload of existing one, in my experience the issue main issue is the amount and length of review cycles. It's better now than a few years ago thanks to the PR managers.

On 03/01/2023 12:25 PM, Dukc wrote: > On Sunday, 25 December 2022 at 18:16:59 UTC, Richard (Rikki) Andrew Cattermole wrote: >> We can't add things to Phobos because of leadership. In the last 10 years we have only added allocators and loggers to experimental, and only one of them was complete enough to come out of experimental. It has nothing at all to do with some flag why we can't put things in. It has always been a nightmare. > > I guess that's true if you're adding an entire module. When adding an individual function, or a new overload of existing one, in my experience the issue main issue is the amount and length of review cycles. It's better now than a few years ago thanks to the PR managers. Yes absolutely. Razvan has been doing an amazing job! I do hope he gets to become fully paid up full time at some point, because he does bring to the table a lot of process improvements. The problem now is a much higher level one, medium to long term, rather than short term. For Phobos anyway.

January 05, 2023

Re: Send me your list of D gripes and wishes

Posted by John Colvin
in reply to Joseph Rushton Wakeling

Permalink

John Colvin

Posted in reply to Joseph Rushton Wakeling

Permalink

On Sunday, 1 January 2023 at 08:53:58 UTC, Joseph Rushton Wakeling wrote:

On Saturday, 31 December 2022 at 18:25:38 UTC, Walter Bright wrote:

True enough. I adjust the style based on what the particular sequence of code is doing. I doubt "good style" can be mandated. Note that D's style checker has a number of knobs that turn various checks off, and this is used in its checking of the compiler source files.

Clearly we are in sympathy here :-) But I've had to recognize that not everyone agrees, and that people have well-motivated reasons for wanting to try to automate the code style. Typically, they want to be able to put 100% focus on what the program is doing and not have to discuss, debate, or spend time manually tweaking stylistic aspects.

This is key. Velocity is increased by decoupling arguments about formatting (go complain to the autoformat authors) from progress on functionality.

The web of dependencies in the software development process (hard like "can't merge B until we merge A" and soft like "I really want to focus on A and not get distracted by B") is complicated and large and mostly hidden. Anything that slows down anything will slow down something else which slows down everything.

Slightly uglier code, but no PR ever waits a day on formatting complaints, no reviewer energy ever goes in to explaining formatting? It's a no-brainer.

Forums