Hi All!
First of all, I want to thank you all for participating in discussion.
So, last week besides doing what was planned(completing file server
interface with existing library functions), I forked the project, and
created read-only version of gitfs with following layout:
$ mount {myfs repo} mntpt
$ cd mntpt
$ ls
drwxrxrx M 4 manzur manzur 314 Jun 25 20:22 master
drwxrxrx M 4 manzur manzur 314 Jun 25 20:22 rdonlygitfs
Each of the above entry is a branch, and each of them contains
following files:
tree - snapshot of the file tree when commit was executed. You can
browse it and read files from it.
parent1 - first parent of commit
...
parentN - Nth parent of commit
commit_msg - commit message
log - log beginning from this commit(at the moment, it's more like old
rev-tree than like git log)
As you can see, it lacks repository manipulation mechanisms. To add
such features I decided to classify git commands. Basic git commands
like git add, rm, cat-file, show will be expressed via cd, ls, cat,
cp, mv and rm.
Other complicated commands will be executed through ctl file, or by
writing to the file itself, if it's a file-specific command.
Each file under some branch will have two modes: command and textual.
Modes are changed by chgrping to the special value. By default, it's
in textual mode.
Git index will be represented as a directory, so that you can use cp,
rm instead of git add, rm. The same is with stages, each of them is a
directory:
index/ - files that're in stage 0
index/1 - files that're in stage 1
index/2 - files that're in stage 2
Let me explain by example:
$ git checkout branch
#...some editing...
$ git add file1
$ git commit -m "msg"
will be changed to
$ cd branch/tree
#..some editing...
$ cp file1 /mnt/index/
$ cp commit_msg /mnt/index
$ mv /mnt/index COMMIT
So why not to use rm?
Because 'rm -r' firstly removes files under index/, but file removing
under index/ is used for git rm.
The last command commits and moves head of the branch to the new
commit.
NOTICE: Editing a commit of the branch makes it working directory. The
other branches will be read-only until you commit or flush changes.
Merge is done through writing to ctl file. I'm not using `cp branch1
branch2` as gitfs needs to differ merging from editing a file. It's
needed to differ them, because gitfs should be able to represent diff
between conflicting versions.
File-specific log.
You can view file-specific log by:
#we should change to command mode
$ chgrp MAGIC file1
$ echo log > file1
$ cat file1
#Log for the file1
That's it. I hope I could explain my decision.
Thanks,
Manzur
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "Plan
9 Google Summer of Code" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at
http://groups.google.com/group/plan9-gsoc?hl=en
-~----------~----~----~----~------~----~------~--~---
