I rarely compress PDF files for emailing or any other purpose. For emailing, I very occasionally apply 7-Zip, not so much for its superior compression, but instead for its AES-256 encryption. 7-Zip - Wikipedia One reason that I do not generally recommend applying compression to PDF files is that much of their included information is already compressed. For example, images within PDFs are typically already encoded. PDF - Wikipedia Other information (e.g. vector graphics and text) are also typically somewhat compressed. PDF supports dictionaries for replacing redundant chunks of information with pointers to their dictionary definitions. A more generally efficient PDF compression would involve first unraveling them and applying compressions optimized for each element. For example, compressing text string information (e.g. Huffman coding - Wikipedia), then including a decompression dictionary along with formatting rules for rendering that recovered textual information during PDF playback.
The decompression dictionary would be optimized for a specific type of text (i.e. a file containing characters as well as words, like emails or PDFs). In addition, any character string of at least a length equal to its length in bytes in a particular file would be considered important information, and would be stored in the dictionary (so it doesn't have to be extracted or otherwise processed at runtime). It might be an improvement on what is currently being done, but still the case where the only benefit is in saving space. Compression-wise, I always try to use compression. I will often decompress in the background, and then apply compression during playback. However, I don't always compress files that I import into Xcode for editing. I have been told by some users that it is not needed, because when editing an existing PDF file, there is a good chance that you will.