Segfault when FCGI parameters are not provided about iipsrv

[ruven]

Thanks for this. I've updated the documentation on the website and in the README regarding NginX.

I've also uploaded a fix based on yours, which is slightly simpler and which keeps the const assignment to request_string and which avoids creating a separate http_if_modified_since string.

[brianhelba]

@ruven , thanks for the quick response to this.

The reason I created a separate "http_if_modified_since" string variable is to avoid doing a map lookup twice. The performance cost of a string copy is much less than that of a map lookup. Further, if you saw my second commit, brianhelba@2d7592b , it's probably better to check whether the "HTTP_IF_MODIFIED_SINCE" parameter is empty, rather than just checking to see whether it exists at all, which is only possible by looking inside the string. Also, creating a string reference won't work, as it has the side effect of inserting "http_if_modified_since" into the map, which breaks later sections of code.

Finally, while I'm happy to see IIP server being improved, I'd also like to be credited for the code that I write. If you need to modify a pull request, it would be nice if you added a patch of your own on top, instead of copy-pasting portions of else's code and committing it with you listed as the Git author. Alternatively, you can comment on the PR and give the author a chance to modify their own work. It's not a huge problem for a small patch like this, but I'd hope that if I submit larger fixes in the future, you will do the courtesy of ensuring that contributors are credited with Git authorship.

[ruven]

Sorry Brian, I did thank and reference you in the commit, but it's true it, in effect, stripped your git author tag.

You're right about the side-effect of inserting an empty "http_if_modified_since" into the map. If you want to resubmit an update of your memcached patch, I'll be happy to merge that directly.