node in mind, we weren't checking to see if that exit node
was already present earlier in the circuit. oops.


svn:r5357

svn:r5356

also note two spec things that need more explanation.


svn:r5355

just confuse our users when they upgrade.


svn:r5354

svn:r5353

maybe change our approach.


svn:r5352

svn:r5351

svn:r5350

On directory servers, old_routers was wasting hundreds of bytes per superseded router descriptor.  Roll the signed descriptor info and identifying info into a cache_info struct, and use only that for old_routers.


svn:r5349

svn:r5348

svn:r5347

svn:r5345

svn:r5344

svn:r5343

svn:r5342

svn:r5341

makes me uncomfortable.


svn:r5340

svn:r5339

svn:r5338

tor_listdir() checks for this too.


svn:r5337

svn:r5336

svn:r5335

if some of our packages statically link any openssl libs.


svn:r5334

125000 circuit_t's after it had been up for a few weeks, which translates
to 20+ megs of wasted space.


svn:r5333

be more fair on loaded Tor servers, and it might resolve our
Windows crash bug. It might also slow things down.


svn:r5332

svn:r5331

svn:r5330

svn:r5329

svn:r5328

svn:r5327

Fix a fun bug that was probably causing unnecessary downloads, and that coupld possibly have caused some segfaults: When post-processing a split fingerprint URL, we were trying to base16_decode() entries already in the fingerprint list, failing, and removing them.  Ow.


svn:r5326

svn:r5325

svn:r5324

Start making directory caches retain old routerinfo_t.  The code to remove old ones is definitely some textbook C problem.


svn:r5323

svn:r5322

clean up OpenSSL's toys when it's done playing.  (Why isn't there an
OpenSSL_free_everything() function?)


svn:r5321

svn:r5320

svn:r5319

svn:r5318

svn:r5317