On 22 April 2018 at 09:51, Jan Iversen <jancasacon...@gmail.com> wrote: > > >> On 22 Apr 2018, at 09:43, sebb <seb...@gmail.com> wrote: >> >> On 20 April 2018 at 12:11, <j...@apache.org> wrote: >>> Hi >>> >>> I have given some thought to the proposal by Sebb that replaces my proposal. >>> >>> There have been on/off talks over a long time about simplifying the >>> maintenance, common for the talks are the wish to only maintain a single >>> file, preferable json format. >> >> Why is a single file necessary? >> >>> The proposal from Sebb have a lot of good qualities, but the latest >>> suggestion is a file pr project combined with yaml files, this is far from >>> the original wish and something I cannot support. >> >> Why not? > Because it is multiple files instead of 1 file.
Huh? Unless you are updating multiple projects, it is only one file. > You need to edit the .md, which eventually will lead to different > content/look&feel project files (with content I mean different classes of > content, like some will have related projects others will not even have the > field…also with .md it is possible to reorder the information). How is that different from JSON data? JSON does not have mandatory attributes, nor does it insist on a particular order. And what does the order matter? >> >>> I am also a bit concerned about having a bot running for something that >>> changes 3-4 times pr year, it seem like a waste of Infra resources. >> >> This same bot is used for lots of sites. >> >>> It is important that the maintenance can be carried out easily and it is >>> important that the look/feel of the site stays identical e.g. old urls must >>> remain available. >> >> Agreed. >> >>> Adding a struct in a json file secures easy maintenance, as in the original >>> proposal. >> >> I disagree that JSON is the way to go. >> >>> Please let us not complicate things. >> >> JSON is complicated to use with anything other than small amounts of >> textual data. > Not really, at least not from my pow, and after all I just converted all > projects to json, that should count for some experience. As I wrote elsewhere, not all the information in the existing XML files has been transferred. It was when I started looking at how to do that with JSON that I started to think that JSON is not the right format. Have a look at how to transfer the additional information from some of the following: Abdera AxKit Beehive Crimson DeviceMap Excalibur HiveMind iBatis ECS ORO Regexp Slide Taglibs (this has a lot of info) Muse OJB Shale Whirr Xang XML XMLBeans Some of that info could perhaps be added to the description field. But I don't think it's practical for everything. And I don't think it's right to drop the information. >> >>> This started because I wanted to simplify my life and have grown into >>> something bigger. >> >>> Bigger might be better, but is it really needed, and are there somebody >>> willing to do our job (move retired project to the Attic) ? >> >> Personally, I would much rather create/edit a single YAML file per >> project than a large slab of JSON. > YAML would solve the problem of a single file, but for that you need to think > about how to online validate the file, before committing. As a personal > opinion I find YAML with its less restrictive format a lot more error prone. > > > rgds > Jan I. >> >>> rgds >>> Jan I. >