svn:r5948

svn:r5947

svn:r5946

svn:r5945

svn:r5944

svn:r5943

svn:r5942

svn:r5941

svn:r5940

svn:r5939

svn:r5938

svn:r5937

other meaning in this context that are confusing.


svn:r5936

svn:r5935

svn:r5934

Document ConnLimit, which is the opposite.


svn:r5933

svn:r5932

sometimes decrement conn->n_circuits even when there was no circuit
originally. This caused conn->n_circuits to go negative.

We noticed this because we were checking if connections can be
closed based on conn->n_circuits == 0, so we were never closing
any connection that had ever had a circuit on it.


svn:r5931

svn:r5930

svn:r5929

need to do it by hand still.
next version, we swear.


svn:r5928

Move "sort list of versions" logic into routerparse.c; make version-checking code say which versions it would have accepted. (not tested.)


svn:r5927

svn:r5926

svn:r5925

svn:r5924

svn:r5923

svn:r5922

svn:r5921

svn:r5920

svn:r5919

svn:r5918

svn:r5917

to think about downloading more server descriptors. this way
we don't have a 10 second pause in initial bootstrapping.


svn:r5916

svn:r5915

svn:r5914

existence of descriptors older than a day.

we still cache them in case anybody asks, but we don't actively
mention them.


svn:r5913

well, forever. i imagine this just keeps growing in size.


svn:r5912

svn:r5911

less often. not critical but might as well stay in sync with stable.


svn:r5910

but i agree that we don't need to worry about them for rendezvousing.
what we really meant was collision attacks. i think?


svn:r5909