Trellis Release Cycles, Testing and HEAD control

This isn’t an issue, more of a curiosity on the release cycles or approaches on changes to Trellis, and if more can be done for Trellis users in providing a stable process when it comes to configuring new projects.

Recently I’ve been configuring many new projects. And due to the activity on Trellis I’ve come across a lot of bugs which have either taken a while to be flagged/fixed or not documented. This has meant that using Trellis has had a lot of pitfalls in the recent weeks, more so than the other elements of the stack like Bedrock and Sage. In comparison those members of the stack seem a lot more structured and tested.

The is not a dig at the Trellis contributors, I follow your change history, the open issues and I’m a very silent yet a very appreciative member of the community. Thank you all while I’m here!

There’s a dated post here which asks for the advice on whether to use a versioned release of Trellis or the HEAD version: Should we use a specific release of Trellis, or HEAD?

The advice given was generally that HEAD on average is stabler than the versioned releases. Backing up this notion of using HEAD, documentation is also generally reflective of HEAD changes and not that of the versioned Trellis.

However, in HEAD, there are currently changes marked as “BREAKING”. More importantly (for me) multisite tests and updates would appear to get tested less before changes are merged/committed to HEAD.

With the above in mind, and imitating using Trellis as a new user following documentation in the past couple of weeks has been troublesome. Not only has documentation been out of sync in some changes with HEAD - yet updated with others. More importantly, without looking through the commit history and following the progress it’s very difficult to figure out if any bugs you run into are localised or due to bugs within recent changes in Trellis.

This leaves both existing Trellis users and new users with a lot of issues and I feel detracts from the main point of Trellis which is an environment setup in a couple of command lines. Granted, I see the argument for having the most upto date changes and this opens up opportunity for users to flag bugs and test these new changes for the good of the community. However, if you’re coming to grab Trellis and use it on the day, at work, or whatever, this experience of bugs could really leave you disgruntled, more so than with Bedrock and Sage.

I’m just wondering what the advice is for using Trellis at the moment? Is grabbing HEAD still advised or should we grab the versioned copy of Trellis while these major updates to Trellis are taking place?

I think if HEAD is the primary branch on the GH repo both documentation and the GH readme should state that experimental, breaking or untested changes have been made to HEAD and advise users to take a versioned branch instead. Does anyone agree?

If so then shouldn’t documentation also be versioned as opposed to almost forcing a user who is following the documentation, to take a HEAD copy with known bugs making Trellis unusable for them on the day? The alternative of course, is that a user grabs the stable version - 0.9.7 for example, and then the documentation (which of course is intended for HEAD) is then defunct…

1 Like

See for more discussion on this