4
4
Don't really need the hashes of the current versions - just knowing
5
5
whether they've changed or not will generally be enough - and just the
6
6
mtime and ctime of a point in time may be enough?
12
There are currently 4 levels that state can have.
15
The actual content blocks have not been read at all.
16
2. IN_MEMORY_UNMODIFIED
17
The content blocks have been read and are available for use. They have
18
not been changed at all versus what was written on disk when we read
20
3. IN_MEMORY_HASH_MODIFIED
21
We have updated the in-memory state, but only to record the
22
sha1/symlink target value and the stat value that means this
23
information is 'fresh'.
25
We have updated an actual record. (Parent lists, added a new file,
26
deleted something, etc.) In this state, we must always write out the
27
dirstate, or some user action will be lost.
30
IN_MEMORY_HASH_MODIFIED
31
~~~~~~~~~~~~~~~~~~~~~~~
33
This state is a bit special, so deserves its own topic. If we are
34
IN_MEMORY_HASH_MODIFIED, we only write out the dirstate if enough records
35
have been updated. The idea is that if we would save future I/O by writing
36
an updated dirstate, then we should do so. The threshold for this is set
37
by "worth_saving_limit". The default is that at least 10 entries must be
38
updated in order to consider the dirstate file worth updating.
40
Going one step further, newly added files, symlinks, and directory entries
41
updates are treated specially. We know that we will always stat all
42
entries in the tree so that we can observe *if* they have changed. In the
43
case of directories, all the information we know about them is just from
44
that stat value. There is no extra content to read. So an update directory
45
entry doesn't cause us to update to IN_MEMORY_HASH_MODIFIED. However, if
46
there are other modifications worth saving, we will go ahead and save the
47
directory entry update at the same time.
49
Similarly, symlink targets are commonly stored in the inode entry
50
directly. So once we have stat'ed the symlink, we already have its target
51
information in memory. The one caveat is if we used to think an object was
52
a file, and it became a directory or symlink, then we will treat it as
55
In the case of newly added files, we never have to read their content to
56
know that they are different from the basis tree. So saving the updated
57
information also won't save a future read.
60
.. vim: ft=rst tw=74 et
added
removed
doc/developers/dirstate.txt
