Crawl package namespaces for CSS/JS deps metadata

[rpkyle]

As originally noted in #50, the current strategy of crawling the Dash layout to identify packages that supply CSS and JavaScript dependencies is potentially problematic.

If the initial layout does not contain any components from a given library, their associated assets will not be loaded. This PR proposes to modify the dependency resolution process somewhat:

• retrieve the namespace information for each currently attached (loaded) package
• subset the list of objects using regex such that only functions whose name matches .dash*_js_metadata are retained (this omits the Dash for R internal metadata, to prevent loading these assets twice)
• as before, retrieve the contents of the metadata function and append to the dependency list for loading

Thanks to @sacul-git for reporting the initial manifestation of this issue, and for supplying a reproducible example for testing purposes.

Closes#50.

@alexcjohnson@chriddyp

[rpkyle]

Initial feedback from users has been positive; the fix appears to work as intended.

[alexcjohnson]

Alright! 💃

[rpkyle]

As originally noted in #50, the current strategy of crawling the Dash layout to identify packages that supply CSS and JavaScript dependencies is potentially problematic.

If the initial layout does not contain any components from a given library, their associated assets will not be loaded. This PR proposes to modify the dependency resolution process somewhat:

• retrieve the namespace information for each currently attached (loaded) package
• subset the list of objects using regex such that only functions whose name matches .dash*_js_metadata are retained (this omits the Dash for R internal metadata, to prevent loading these assets twice)
• as before, retrieve the contents of the metadata function and append to the dependency list for loading

Thanks to @sacul-git for reporting the initial manifestation of this issue, and for supplying a reproducible example for testing purposes.

Closes#50.

@alexcjohnson@chriddyp

[rpkyle]

Initial feedback from users has been positive; the fix appears to work as intended.

[alexcjohnson]

Alright! 💃

[rpkyle]

As originally noted in #50, the current strategy of crawling the Dash layout to identify packages that supply CSS and JavaScript dependencies is potentially problematic.

If the initial layout does not contain any components from a given library, their associated assets will not be loaded. This PR proposes to modify the dependency resolution process somewhat:

• retrieve the namespace information for each currently attached (loaded) package
• subset the list of objects using regex such that only functions whose name matches .dash*_js_metadata are retained (this omits the Dash for R internal metadata, to prevent loading these assets twice)
• as before, retrieve the contents of the metadata function and append to the dependency list for loading

Thanks to @sacul-git for reporting the initial manifestation of this issue, and for supplying a reproducible example for testing purposes.

Closes#50.

@alexcjohnson@chriddyp

[rpkyle]

Initial feedback from users has been positive; the fix appears to work as intended.

[alexcjohnson]

Alright! 💃

[rpkyle]

As originally noted in #50, the current strategy of crawling the Dash layout to identify packages that supply CSS and JavaScript dependencies is potentially problematic.

If the initial layout does not contain any components from a given library, their associated assets will not be loaded. This PR proposes to modify the dependency resolution process somewhat:

• retrieve the namespace information for each currently attached (loaded) package
• subset the list of objects using regex such that only functions whose name matches .dash*_js_metadata are retained (this omits the Dash for R internal metadata, to prevent loading these assets twice)
• as before, retrieve the contents of the metadata function and append to the dependency list for loading

Thanks to @sacul-git for reporting the initial manifestation of this issue, and for supplying a reproducible example for testing purposes.

Closes#50.

@alexcjohnson@chriddyp

[rpkyle]

Initial feedback from users has been positive; the fix appears to work as intended.

[alexcjohnson]

Alright! 💃

[rpkyle]

As originally noted in #50, the current strategy of crawling the Dash layout to identify packages that supply CSS and JavaScript dependencies is potentially problematic.

If the initial layout does not contain any components from a given library, their associated assets will not be loaded. This PR proposes to modify the dependency resolution process somewhat:

• retrieve the namespace information for each currently attached (loaded) package
• subset the list of objects using regex such that only functions whose name matches .dash*_js_metadata are retained (this omits the Dash for R internal metadata, to prevent loading these assets twice)
• as before, retrieve the contents of the metadata function and append to the dependency list for loading

Thanks to @sacul-git for reporting the initial manifestation of this issue, and for supplying a reproducible example for testing purposes.

Closes#50.

@alexcjohnson@chriddyp

[rpkyle]

Initial feedback from users has been positive; the fix appears to work as intended.

[alexcjohnson]

Alright! 💃

[rpkyle]

As originally noted in #50, the current strategy of crawling the Dash layout to identify packages that supply CSS and JavaScript dependencies is potentially problematic.

If the initial layout does not contain any components from a given library, their associated assets will not be loaded. This PR proposes to modify the dependency resolution process somewhat:

• retrieve the namespace information for each currently attached (loaded) package
• subset the list of objects using regex such that only functions whose name matches .dash*_js_metadata are retained (this omits the Dash for R internal metadata, to prevent loading these assets twice)
• as before, retrieve the contents of the metadata function and append to the dependency list for loading

Thanks to @sacul-git for reporting the initial manifestation of this issue, and for supplying a reproducible example for testing purposes.

Closes#50.

@alexcjohnson@chriddyp

[rpkyle]

Initial feedback from users has been positive; the fix appears to work as intended.

[alexcjohnson]

Alright! 💃

[rpkyle]

As originally noted in #50, the current strategy of crawling the Dash layout to identify packages that supply CSS and JavaScript dependencies is potentially problematic.

If the initial layout does not contain any components from a given library, their associated assets will not be loaded. This PR proposes to modify the dependency resolution process somewhat:

• retrieve the namespace information for each currently attached (loaded) package
• subset the list of objects using regex such that only functions whose name matches .dash*_js_metadata are retained (this omits the Dash for R internal metadata, to prevent loading these assets twice)
• as before, retrieve the contents of the metadata function and append to the dependency list for loading

Thanks to @sacul-git for reporting the initial manifestation of this issue, and for supplying a reproducible example for testing purposes.

Closes#50.

@alexcjohnson@chriddyp

[rpkyle]

Initial feedback from users has been positive; the fix appears to work as intended.

[alexcjohnson]

Alright! 💃

[rpkyle]

As originally noted in #50, the current strategy of crawling the Dash layout to identify packages that supply CSS and JavaScript dependencies is potentially problematic.

If the initial layout does not contain any components from a given library, their associated assets will not be loaded. This PR proposes to modify the dependency resolution process somewhat:

• retrieve the namespace information for each currently attached (loaded) package
• subset the list of objects using regex such that only functions whose name matches .dash*_js_metadata are retained (this omits the Dash for R internal metadata, to prevent loading these assets twice)
• as before, retrieve the contents of the metadata function and append to the dependency list for loading

Thanks to @sacul-git for reporting the initial manifestation of this issue, and for supplying a reproducible example for testing purposes.

Closes#50.

@alexcjohnson@chriddyp

[rpkyle]

Initial feedback from users has been positive; the fix appears to work as intended.

[alexcjohnson]

Alright! 💃