While this is not necessarily the perfect way, here are some basics for a good workflow:

• Shoot in RAW: this format allows to restore highlights and to boost shadows above a larger range than jpeg. Furthermore the white balance can be adjusted later without losses. The color space you choose does not matter, as this captures the raw* values from the sensor readout.
  *with some model specific processing

• Develop / Process your images nondestructive. Programs like Lightroom and Darktable wont change the RAW file but use a history of changesets that will be applied on the preview and the exported (jpeg) file.

• While developing choose your target format / profile you want to export to. This is called softproofing. For a web presentation or digital sharing the target would be set to sRGB. For printing one needs the printer profile, which is either provided by the manufacturer or the printing service used. These profiles show the color space a printer is capable of reproducing. Make sure that the amount of clipped highlights and shadows can be tolerated. Also watch for differences in color gradients ( most common: blue sky with flat color gradient i.e. only small changes between each pixel). This can lead to artifacts in the final picture.

• Export your picture to a format and color space that fits to the kind of publishing. Publishing online would call for jpeg in sRGB. Good printing services will state, which format and color space they accept or prefer. The best compromise IMO is to use jpeg with low compression (high quality setting) and the largest color space accepted by the printing service. This could be AdobeRGB or ProPhoto RGB. Note that the color values of your image have to fit into the color space of the printer. A larger embedded color space allows you to use all or most color values a printer can reproduce. sRGB usually is a bad match for the color spaces of printers.