digitalmars.D - Tools should use the review process
- Jacob Carlborg (11/11) Dec 28 2013 Since we stared with a review process for adding new modules to Phobos
- Andrei Alexandrescu (16/25) Dec 28 2013 That would, of course, be great.
- Martin Nowak (2/5) Dec 28 2013 Yep
- Jacob Carlborg (17/31) Dec 28 2013 I know and that sucks. But is that a reason to not have the same
- H. S. Teoh (18/27) Dec 28 2013 [...]
- Jacob Carlborg (4/16) Dec 29 2013 Well said. It's most important to have a review for the end user tools.
- Andrei Alexandrescu (4/11) Dec 28 2013 You know how it works - great tools for everyone originate as great
- Jacob Carlborg (6/8) Dec 29 2013 For some tools it's more obvious that they are benefit to both end users...
- Dicebot (6/9) Dec 29 2013 There is a very practical concern in separation - end-user tools
Since we stared with a review process for adding new modules to Phobos around eight new tools have been added without any review process similar to the one used for Phobos modules. They were only reviewed in pull requests. Some of these are minor tools/scripts but there are larger tools as well, like RDMD (added before the release process) and DustMire (added after the release process). I think that new tools should use the same review process just as new modules for Phobos do. What do you think? -- /Jacob Carlborg
Dec 28 2013
On 12/28/13 6:16 AM, Jacob Carlborg wrote:Since we stared with a review process for adding new modules to Phobos around eight new tools have been added without any review process similar to the one used for Phobos modules. They were only reviewed in pull requests. Some of these are minor tools/scripts but there are larger tools as well, like RDMD (added before the release process) and DustMire (added after the release process). I think that new tools should use the same review process just as new modules for Phobos do. What do you think?That would, of course, be great. I can't stop mentioning again that we are having a severe deflation of reviewers. We are currently treating our contributors the worst way possible - with indifference. The oldest pull request is two years old: https://github.com/D-Programming-Language/dmd/pull/398. There is activity on the new pull requests, but many old pull requests are just left unreviewed. This is especially painful for new or casual contributors, who make an attempt and are discouraged by the lack of care. Also, due to there being fewer attentive reviewers than contributors, some requests are pulled without sufficient review. That makes me worried about adding another repo to the workload of reviewers - it means spreading an already insufficient resource even thinner. Of course the right solution to this is more review participation. Andrei
Dec 28 2013
On 12/28/2013 06:28 PM, Andrei Alexandrescu wrote:I can't stop mentioning again that we are having a severe deflation of reviewers. We are currently treating our contributors the worst way possible - with indifference.Yep
Dec 28 2013
On 2013-12-28 18:28, Andrei Alexandrescu wrote:That would, of course, be great. I can't stop mentioning again that we are having a severe deflation of reviewers. We are currently treating our contributors the worst way possible - with indifference.I know and that sucks. But is that a reason to not have the same standard as for new Phobos modules?The oldest pull request is two years old: https://github.com/D-Programming-Language/dmd/pull/398. There is activity on the new pull requests, but many old pull requests are just left unreviewed. This is especially painful for new or casual contributors, who make an attempt and are discouraged by the lack of care. Also, due to there being fewer attentive reviewers than contributors, some requests are pulled without sufficient review.I agree. There are even older pull requests that are closed by the contributor due to lack of interest, either from the reviewers or/and contributors side https://github.com/D-Programming-Language/dmd/pull/3That makes me worried about adding another repo to the workload of reviewers - it means spreading an already insufficient resource even thinner. Of course the right solution to this is more review participation.I know and agree, but I don't want the quality to be any less then what it is in the Phobos modules. One think I noticed in the tools repository is that some tools are more focused towards the core team, like changed.d and some tools are for all developers like RDMD. The most important tools are of course those aimed to be used by all developers. Perhaps we can find a middle ground here. I'm thinking that Dub is a possible candidate for inclusion in the tools repository. That would be a perfect example for a tool aimed at to all developers and should go through a review process. -- /Jacob Carlborg
Dec 28 2013
On Sat, Dec 28, 2013 at 09:03:04PM +0100, Jacob Carlborg wrote:On 2013-12-28 18:28, Andrei Alexandrescu wrote:[...]One think I noticed in the tools repository is that some tools are more focused towards the core team, like changed.d and some tools are for all developers like RDMD. The most important tools are of course those aimed to be used by all developers. Perhaps we can find a middle ground here. I'm thinking that Dub is a possible candidate for inclusion in the tools repository. That would be a perfect example for a tool aimed at to all developers and should go through a review process.[...] IMNSHO, I think the tools repo needs a reorg. Currently we have trivial things like 'changed' or 'findtags' or 'latest-tag', which are completely useless to everyone except the core devs (you probably don't even know what they're supposed to do unless you read the source code), and then we have 'dustmite' and 'rdmd' that generally applicable to whoever uses D. I think this needs to be split into two repos, one for production-quality tools that are meant for the end-user, and one for miscellaneous quick-hack scripts meant for internal use by the core devs. The current situation of dumping random stuff into tools that most people may or may not care for, 'cos we don't know where else to put them, only makes a laughing stock of D. T -- Маленькие детки - маленькие бедки.
Dec 28 2013
On 2013-12-28 21:28, H. S. Teoh wrote:IMNSHO, I think the tools repo needs a reorg. Currently we have trivial things like 'changed' or 'findtags' or 'latest-tag', which are completely useless to everyone except the core devs (you probably don't even know what they're supposed to do unless you read the source code), and then we have 'dustmite' and 'rdmd' that generally applicable to whoever uses D. I think this needs to be split into two repos, one for production-quality tools that are meant for the end-user, and one for miscellaneous quick-hack scripts meant for internal use by the core devs. The current situation of dumping random stuff into tools that most people may or may not care for, 'cos we don't know where else to put them, only makes a laughing stock of D.Well said. It's most important to have a review for the end user tools. -- /Jacob Carlborg
Dec 29 2013
On 12/28/13 12:03 PM, Jacob Carlborg wrote:One think I noticed in the tools repository is that some tools are more focused towards the core team, like changed.d and some tools are for all developers like RDMD. The most important tools are of course those aimed to be used by all developers. Perhaps we can find a middle ground here. I'm thinking that Dub is a possible candidate for inclusion in the tools repository. That would be a perfect example for a tool aimed at to all developers and should go through a review process.You know how it works - great tools for everyone originate as great tools for developers. Andrei
Dec 28 2013
On 2013-12-28 21:44, Andrei Alexandrescu wrote:You know how it works - great tools for everyone originate as great tools for developers.For some tools it's more obvious that they are benefit to both end users and core developers. See the post by H. S. Teoh: http://forum.dlang.org/thread/l9mmfd$1tte$1 digitalmars.com#post-mailman.85.1388262606.15871.digitalmars-d:40puremagic.com -- /Jacob Carlborg
Dec 29 2013
On Saturday, 28 December 2013 at 20:44:10 UTC, Andrei Alexandrescu wrote:You know how it works - great tools for everyone originate as great tools for developers. AndreiThere is a very practical concern in separation - end-user tools should be packages as part of compiler distribution. Right now I chose tools to add to Arch Linux `dlang-tools` package based on my own judgement. I'd prefer it to be standard obviously.
Dec 29 2013