go-module | : no_entry : This project

I am not good at using the debug tools, so just stepped through the code. (Appreciate the suggestion of good tutorials!)

First things first: When in Doubt, Try to Upgrade Your Software Packages This is the blogdown package creator's advice. Checked.

1. Launching the addin, blogdown:::new_post_addin is called.
2. This calls new_post.R. The source code can be find at: https://github.com/rstudio/blogdown/blob/master/inst/scripts/new_post.R
3. blogdown::new_post() is called at the end, with file parameter from the updated input text field, which is in the case of the question: "post\2020-11-04-how-this-site-was-created\index.en.md"
4. new_post() function (in hugo.R) calls new_content() with the third argument, open = FALSE, which means it will not open the file, just overwrite the value of the file variable (the path hereafter). At the end of new_post() the file should be opened: this is where the error occurs (by trying to open the file at the wrong place, a wrong path). This means that something bad has to happen in new_content().
5. new_content() modify the path with content_file(). Also this is the step, where the file is actually created.
6. content_file() modify the path by adding a prefix to it generated by get_config().
7. get_config() tries to extract the value of a field in the configuration, in the case of Hugo: try to find contentDir. If this results in NULL (there's no such list item), then see the other possibility and lastly return the default value (which is in the case of Hugo: content). And that's it! RStudio tries to open a file in the content directory, not in a language subdirectory!
8. That means that the last option is returned in the row of %n% (which is imported from knitr and stands for: if (is.null(x)) y else x). That means that the previous arguments returned NULL, which means contentDir cannot be found in config variable. The default value of config is set by config = load_config(). This uses find_config() to find the config file to parse.
9. find_config() uses config_files() to set the value in case of Hugo: c('config.toml', 'config.yaml'). But my setting is in a subdirectory: in the /config/default_/languages.toml file! Oh, another Academic theme woe...

To conclude this: blogdown is currently loads only the config.toml in the root directory to check the contentDir value. The Academic, now Wowchemy Hugo theme however keep config files in the /config/default_/ directory also, where the languages.toml contains the needed value.

To keep or not to keep Academic theme? Stick with blogdown or not? Maybe the config files should be merged into one config.toml, but couldn't find hint or examples for Academic on the web.

Edit: OK, the root of the problem has been found, but the actual error which raised the error message is in connection with hugo_convert_one(), as the traceback suggests in the question. This was the first (and last) try to open the file of the wrong path.

You'll find it a lot easier if you have your imports relative to the project directory (the one that contains manage.py)

Following the Edit your site on your PC docs on the Academic theme's webpage:

Before downloading your site, lets first install Hugo Extended and its prerequisites.

On Windows: Git, OpenSSH and Go have to be installed beside Hugo Extended. Because I already have Git, I did not install it, only OpenSSH and Hugo Extended via Scope (and beforehand Powershell 5 to install Scope). Go was installed via the executable from the Go documentaion site.

After this, Academic theme can be deployed with the RStudio method specified in the question. Maybe the blogdown book need to be updated.

Note: also, in the referenced document:

Convert an old Academic Kickstarter site If you have an existing site based on the Academic Kickstarter Template that was created before 3rd September 2020, it may need converting to use Hugo’s new modular system.

The post_save signal is sent by the model Entry after a save, but according to this source, the signal does not include the model changes in its update_fields parameter. These must be manually inserted in the override of the function where the save() happens; which defeats the purpose. I initially assumed that the signal would automatically include this information.

It is possible since every step on Cloud Build is executed in a docker container, but the image gcr.io/cloud-builders/gcloud only has a minimum installation of the gcloud components, before start the emulator your need to install the pubsub emulator via gcloud command

I had the same problem, and resolved it by editing my config.toml file:

Check that your theme is set correctly (mine was):

Your module is called example.com/libGo and the package within this module is called libGo. Therfore the full package name would be example.com/libGo/libGo.

You would need to adjust the import or the module name. When you adjust the import to example.com/libGo/libGo, the module name must stay example.com/libGo. When you adjust the module name to example.com, the import name must stay example.com/libGo. Adding the package name to both ones puts you in the same situation as before.

The vendor file was being used, it just didn't seem like it because although it was not re-downloading all modules on build it was re-building them on every build. The issue appears to be trying to use the builder pattern, I have altered my development compose file to handle everything in the compose yaml and will reserve the builder pattern Dockerfile for production (where it only really matters anyway).

Now using the following my development builds are way faster and don't appear to recompile every module on every build:

docker-compose.yaml

I came across the answer (by a Google engineer working on this product) here as well as in Error when trying to deploy Google cloud function in Go 1.11 using go modules. The key point is that gcloud functions deploy only copies the contents of the directory in which it is run into the Cloud Function's 'context', so the vendor/ directory has to be in that directory. At the same time, I noticed that the package cannot be main, so I took the somewhat unusual step of removing the go.mod from the repo's root directory and running go mod init and go mod vendor in the helloworld directory. Now it works: