You asked for the long answer, I'll give it to you so that this issue can be put to bed once and for all. Once upon a time we used subjects as a filename, but switched away from that approach years ago due to the problems that ensued as a result. Subjects tend to contain characters in them that do not comply with the acceptable character requirements imposed by the various file systems. When adding a subject as a filename, one has to account for the fact that there are duplicate subjects. The system, therefore, needs to check whether the file exists or not prior to writing the file. Believe it or not, this extra exists check, slows down the export process because an additional I/O operation must be performed on export of each document. We have some customers who perform exports of 100k emails at a time. Performance is critical for them. Furthermore, when writing a file, it becomes necessary to strip out the illegal characters from the filename. This issue is compounded by the fact that different filesystems accept different character sets in a filename. In many cases, after stripping the combined intersection of all illegal characters across all filesystems, one may be left with a husk of a filename that increases the likelihood of filename conflicts with other files already written to disk. To avoid conflicts, it is then necessary, append a numeric sequential suffix to the filename. Ultimately, the suffixes can get into the thousands. You end up with file like, "123233.eml", "
123234 .eml", etc. Totally meaningless. In short, for practical reasons, we don't use subjects as a filename.