add support for streaming clone.
existing clone code uses pull to get changes from remote repo. is very
slow, uses lots of memory and cpu.
new clone code has server write file data straight to client, client
writes file data straight to disk. memory and cpu used are very low,
clone is much faster over lan.
new client can still clone with pull, can still clone from older servers.
new server can still serve older clients.
#!/bin/sh
mkdir t
cd t
hg init
mkdir -p beans
for b in kidney navy turtle borlotti black pinto; do
echo $b > beans/$b
done
mkdir -p mammals/Procyonidae
for m in cacomistle coatimundi raccoon; do
echo $m > mammals/Procyonidae/$m
done
echo skunk > mammals/skunk
echo fennel > fennel
echo fenugreek > fenugreek
echo fiddlehead > fiddlehead
echo glob:glob > glob:glob
hg addremove
hg commit -m "commit #0" -d "1000000 0"
hg debugwalk
cd mammals
hg debugwalk .
hg debugwalk Procyonidae
cd Procyonidae
hg debugwalk .
hg debugwalk ..
cd ..
hg debugwalk ../beans
hg debugwalk .
hg debugwalk .hg
hg debugwalk ../.hg
cd ..
hg debugwalk -Ibeans
hg debugwalk 'glob:mammals/../beans/b*'
hg debugwalk '-X*/Procyonidae' mammals
hg debugwalk path:mammals
hg debugwalk ..
hg debugwalk beans/../..
hg debugwalk .hg
hg debugwalk beans/../.hg
hg debugwalk beans/../.hg/data
hg debugwalk beans/.hg
# Don't know how to test absolute paths without always getting a false
# error.
#hg debugwalk `pwd`/beans
#hg debugwalk `pwd`/..
hg debugwalk glob:\*
hg debugwalk 're:.*[kb]$'
hg debugwalk path:beans/black
hg debugwalk beans 'glob:beans/*'
hg debugwalk 'glob:j*'
hg debugwalk NOEXIST
mkfifo fifo
hg debugwalk fifo
rm fenugreek
hg debugwalk fenugreek
hg rm fenugreek
hg debugwalk fenugreek
touch new
hg debugwalk new