stopgap doesn't load on Windows about aenea

[calmofthestorm]

Stopgap sas the first grammar I wrote. It used to contain basically everything, now it contains almost nothing -- just random launchers. h is the hyper key, and I do not know if that is a meaningful concept on Windows. At any rate nothing in it seems terribly useful on Windows or honestly to anyone not using my window manager config.

One thing that needs to happen at some point is the grammar directory needs to transition from being my grammars that I like to a collection of grammars that may be of use. I'm not quite sure how to go about this but now that several other people are actively contributing to the repository it seems like the right time.

The simplest possibility is probably to reclaim the grammar directory as a place where any grammars can go, and then have something like a grammar_deploy directory that users can copy and or symlink grammars into, and where reload-configuration would pull from.

Another possibility would be for me to create a separate repository for the grammars I wrote that are not likely to be of use to others (except possibly as starting points) and leave in aenea only grammars that are clean and narrowly tailored to a particular goal.

Do you have any suggestions about the right way to coordinate specific grammar development? Essentially I feel I currently my use case has too much influence on the organization, if that makes sense.

[nirvdrum]

It's tricky because everyone has a different set of behaviors they'd like to perform with their own vocabulary. For my part, I'm willing to adapt to others in the name of normalization and so I can get a good experience out of the box.

But the other way to go is not to provide any application-specific grammars. You'd then have a framework for performing common programmer-type translations such as snake-casing and camel-casing. I don't know if that would be preferable or not though.

[calmofthestorm]

I think my ideal system would basically be a package manager -- a large repository of grammars you can choose to or enable or not easily. Perhaps with a default set enabled to cover the case of out of box experience.

I think for now I will move to a structure with grammars_available and grammars_enabled, and let people symlink/add their own modules as desired. In the future I may split it off into its own repository, as dragonfly has done for modules. This also means I won't feel as bad about having modules that only work via proxy -- I would prefer to minimize incompatibility, but I don't have a problem with it in principle if it's something that only makes sense on one platform.

At the end of the day, I feel like there's really no clean way to organize a project like this -- it's not a single piece of software, but a collection of things working together. Feels more like sysadminery than programming if that makes sense :-) if it weren't for copyright law I would simply ship a Windows image with Dragon on it and call it a day:-)

[nirvdrum]

That'd be fine, other symlinks not really existing on Windows. I guess there's junction points, but I don't think explorer has a way to create them. Copy & paste would have to do.

I'm fine with the divide, too. For my part, I'm getting my feet wet trying to set up grammars for Windows apps with Linux equivalents. So nothing I'm writing yet has been tested out on the proxy side of things, although I'd love for them to work that way too.

[calmofthestorm]

We may have to content ourselves with copy/paste on windows; junction points are hard links, not symlinks (a linux symlink is like a windows shortcut, but they behave differently in terms of default behavior when programs access them, so that you can usually drop a symlink in somewhere in place of a real file and programs that don't try to notice won't).

[calmofthestorm]

5922bc7 reorganizes the grammar as described above.