Merge lp:~xaav/loggerhead/export-tarball into lp:loggerhead

[xaav]

This branch **may** accomplish exporting the tarball using chunked transfer encoding. The code all looks to be correct, but I have not tested it, so I would like your opinion.

Thanks!

[Martin Pool (mbp)]

Thanks very much, that'd be a really useful feature to have. Thanks
also for making it optional, because probably some installations would
not want it on.

This looks broadly reasonable -- I'm not deeply familiar with
loggerhead -- but I am very curious why you apparently reimplemented
the export-to-tarball feature. I'd rather reuse the bzr code and if
necessary change it to let it be reused here.

When you say "not tested" do you mean you haven't even run it, or only
that you didn't add automatic tests?

[Robert Collins (lifeless)]

This looks very interesting.

Three isues:

1)
+class TarExporterFileObject(object):
+
+ def __init__(self):
+ self._buffer = ''
+
+ def write(self, str):
+ self._buffer += str
+
+ def get_buffer(self):
+ buffer = self._buffer
+ self._buffer = ''
+ return buffer

This is going to be somewhat inefficient. Try this:

+class TarExporterFileObject(object):
+
+ def __init__(self):
+ self._buffer = []
+
+ def write(self, str):
+ self._buffer.append(str)
+
+ def get_buffer(self):
+ try:
+ return ''.join(self._buffer)
+ finally:
+ self._buffer = []

2) There are no tests for this. the test suite is still pretty new, but its a good idea to test things - in particular in cases like this we need to be fairly confident it will be incremental and not block on the export - I can imagine the wsgi layer buffering everything, for instance. [in fact, I'll lay odds it will unless we fix a few things].

3) The export function is a copy-paste-tweak of the core from bzrlib. This will lead to bugs as that code base evolves - we should instead get a supported function in bzrlib lib that we can import and use.

I'm putting this back to WIP - but its a great start. Please keep at it and just shout if you need pointers.

review: Needs Fixing

[xaav (xaav)]

> 3) The export function is a copy-paste-tweak of the core from bzrlib. This will lead to bugs as
> that code base evolves - we should instead get a supported function in bzrlib lib that we can
> import and use.

Well, there is only one problem with that. According to the WSGI spec, you must return an iterable that will export the blocks. If I were to call the provided function, it would be impossible to break the response into pieces because the provided function would export it all at once. I know that it is a copy and paste tweak, but there is really no way I can inject the 'yield' keyword into the provided function. If you have another suggestion, I would be glad to hear it.

Issue number one I will be glad to fix.

Regarding issue number two, I have not written tests before but I will try my best.

[Robert Collins (lifeless)]

On Wed, Jun 1, 2011 at 9:46 AM, Geoff <email address hidden> wrote:
>> 3) The export function is a copy-paste-tweak of the core from bzrlib. This will lead to bugs as
>> that code base evolves - we should instead get a supported function in bzrlib lib that we can
>> import and use.
>
> Well, there is only one problem with that. According to the WSGI spec, you must return an iterable that will export the blocks. If I were to call the provided function, it would be impossible to break the response into pieces because the provided function would export it all at once. I know that it is a copy and paste tweak, but there is really no way I can inject the 'yield' keyword into the provided function. If you have another suggestion, I would be glad to hear it.

Extract the function in bzrlib into two parts - a generator (what you
have here) and a consumer than consumes it all triggering the writes.

Then we can reuse the generator.

[xaav (xaav)]

> On Wed, Jun 1, 2011 at 9:46 AM, Geoff <email address hidden> wrote:
> >> 3) The export function is a copy-paste-tweak of the core from bzrlib. This
> will lead to bugs as
> >> that code base evolves - we should instead get a supported function in
> bzrlib lib that we can
> >> import and use.
> >
> > Well, there is only one problem with that. According to the WSGI spec, you
> must return an iterable that will export the blocks. If I were to call the
> provided function, it would be impossible to break the response into pieces
> because the provided function would export it all at once. I know that it is a
> copy and paste tweak, but there is really no way I can inject the 'yield'
> keyword into the provided function. If you have another suggestion, I would be
> glad to hear it.
>
> Extract the function in bzrlib into two parts - a generator (what you
> have here) and a consumer than consumes it all triggering the writes.
>
> Then we can reuse the generator.

Okay, I see what you mean.

[xaav (xaav)]

See Bug #791005 for further information on this.

[xaav (xaav)]

Okay, I think this should work, but I haven't tested it yet.

[Martin Pool (mbp)]

Thanks, xaav.

When you say "haven't tested" do you mean just "not written any tests", or "not even been able to run it"?

[xaav (xaav)]

I haven't written any tests (Sorry, I'll do this ASAP.)
I also have not been able to run it because I'm lazy and it requires too much work to get loggerhead running on W1nd0w$.

[xaav (xaav)]

Okay, I've added a simple tarfile test. However, I am not able to run the test to I would appreciate if someone would do that for me and/or try to download a tarball from their browser.

[Gavin Panella (allenap)]

Once I'd set up a virtualenv with the right prerequisites, I got the
following error when running the test suite:

{{{
Traceback (most recent call last):
  ...
  File ".../loggerhead/tests/test_controllers.py", line 8, in <module>
    from loggerhead.apps.branch import BranchWSGIApp
  File ".../loggerhead/apps/branch.py", line 36, in <module>
    from loggerhead.controllers.download_ui import DownloadUI, DownloadTarballUI
  File ".../loggerhead/controllers/download_ui.py", line 29, in <module>
    from loggerhead.exporter import export_tarball
ImportError: cannot import name export_tarball
}}}

After fixing that I got the following error from
TestDownloadTarballUI.test_download_tarball:

{{{
Traceback (most recent call last):
  File "/usr/lib/python2.7/dist-packages/testtools/runtest.py", line 169, in _run_user
    return fn(*args, **kwargs)
  File "/usr/lib/python2.7/dist-packages/testtools/testcase.py", line 499, in _run_test_method
    return self._get_test_method()()
  File ".../loggerhead/tests/test_controllers.py", line 135, in test_download_tarball
    app = self.setUpLoggerhead()
  File ".../loggerhead/tests/test_simple.py", line 47, in setUpLoggerhead
    branch_app = BranchWSGIApp(self.tree.branch, '', **kw).app
AttributeError: 'TestDownloadTarballUI' object has no attribute 'tree'
}}}

Obviously this needs some work.

We've been talking about taking more of a "patch pilot" approach in
Launchpad. That seems to mean that one of the core team - fwiw, I
would be happy to do it - would actively help getting this landed,
rather than just reviewing it. Would you like that, or would you
prefer to iterate on your own?

review: Needs Fixing

[xaav (xaav)]

Sorry, I had been gone. I will be sure and look into this right away!

On Thu, Jun 16, 2011 at 8:56 AM, Gavin Panella
<email address hidden>wrote:

> Review: Needs Fixing
> Once I'd set up a virtualenv with the right prerequisites, I got the
> following error when running the test suite:
>
> {{{
> Traceback (most recent call last):
> ...
> File ".../loggerhead/tests/test_controllers.py", line 8, in <module>
> from loggerhead.apps.branch import BranchWSGIApp
> File ".../loggerhead/apps/branch.py", line 36, in <module>
> from loggerhead.controllers.download_ui import DownloadUI,
> DownloadTarballUI
> File ".../loggerhead/controllers/download_ui.py", line 29, in <module>
> from loggerhead.exporter import export_tarball
> ImportError: cannot import name export_tarball
> }}}
>
> After fixing that I got the following error from
> TestDownloadTarballUI.test_download_tarball:
>
> {{{
> Traceback (most recent call last):
> File "/usr/lib/python2.7/dist-packages/testtools/runtest.py", line 169, in
> _run_user
> return fn(*args, **kwargs)
> File "/usr/lib/python2.7/dist-packages/testtools/testcase.py", line 499,
> in _run_test_method
> return self._get_test_method()()
> File ".../loggerhead/tests/test_controllers.py", line 135, in
> test_download_tarball
> app = self.setUpLoggerhead()
> File ".../loggerhead/tests/test_simple.py", line 47, in setUpLoggerhead
> branch_app = BranchWSGIApp(self.tree.branch, '', **kw).app
> AttributeError: 'TestDownloadTarballUI' object has no attribute 'tree'
> }}}
>
> Obviously this needs some work.
>
> We've been talking about taking more of a "patch pilot" approach in
> Launchpad. That seems to mean that one of the core team - fwiw, I
> would be happy to do it - would actively help getting this landed,
> rather than just reviewing it. Would you like that, or would you
> prefer to iterate on your own?
>
> --
> https://code.launchpad.net/~xaav/loggerhead/export-tarball/+merge/63931
> You are the owner of lp:~xaav/loggerhead/export-tarball.
>

[xaav (xaav)]

We've been talking about taking more of a "patch pilot" approach in
Launchpad. That seems to mean that one of the core team - fwiw, I
would be happy to do it - would actively help getting this landed,
rather than just reviewing it. Would you like that, or would you
prefer to iterate on your own?

That would be great! Any help would be greatly appreciated.

[xaav (xaav)]

Okay, I've fixed some stuff (the tests are still broken), but here is what I'm getting:

Traceback (most recent call last):
  File "C:\Python27\lib\site-packages\paste-1.7.5.1-py2.7.egg\paste\httpserver.py", l
ine 1068, in process_request_in_thread
    self.finish_request(request, client_address)
  File "C:\Python27\lib\SocketServer.py", line 323, in finish_request
    self.RequestHandlerClass(request, client_address, self)
  File "C:\Python27\lib\SocketServer.py", line 639, in __init__
    self.handle()
  File "C:\Python27\lib\site-packages\paste-1.7.5.1-py2.7.egg\paste\httpserver.py", l
ine 442, in handle
    BaseHTTPRequestHandler.handle(self)
  File "C:\Python27\lib\BaseHTTPServer.py", line 337, in handle
    self.handle_one_request()
  File "C:\Python27\lib\site-packages\paste-1.7.5.1-py2.7.egg\paste\httpserver.py", l
ine 437, in handle_one_request
    self.wsgi_execute()
  File "C:\Python27\lib\site-packages\paste-1.7.5.1-py2.7.egg\paste\httpserver.py", l
ine 289, in wsgi_execute
    for chunk in result:
  File "C:\Users\Xaav\workspace\loggerhead\loggerhead\exporter.py", line 31, in e
xport_archive
    for _ in get_export_generator(tree=tree, fileobj=fileobj, format=format):
  File "C:\Python27\lib\site-packages\bzrlib\export\__init__.py", line 112, in get_ex
port_generator
    root = get_root_name(dest)
  File "C:\Python27\lib\site-packages\bzrlib\export\__init__.py", line 174, in get_ro
ot_name
    dest = os.path.basename(dest)
  File "C:\Python27\lib\ntpath.py", line 198, in basename
    return split(p)[1]
  File "C:\Python27\lib\ntpath.py", line 170, in split
    d, p = splitdrive(p)
  File "C:\Python27\lib\ntpath.py", line 125, in splitdrive
    if p[1:2] == ':':
TypeError: 'NoneType' object is not subscriptable

Any help would be appreciated.

[Gavin Panella (allenap)]

Cool. I am busy this week, but I might get to it. If not, next week for sure.

[Vincent Ladeuil (vila)]

Hi,

I almost got the test running with some additional fixes
(available at lp:~vila/loggerhead/export-tarball) only to run
into a bug in bzr itself (I think you should be able to fix that
one ;).

Note that your code requires bzr >= 2.4 (launchpad only runs
2.3.3 so far) so we'll need some support from the lp guys to
deploy a more recent version there.

Summary of my fixes:

- you need to create a branch (with some content even) before
  being able to call

         app = self.setUpLoggerhead()

  So I've added a setUp method for your class to do that. You
  probably want to add *more* tests to check that you get a valid
  tarball with the expected content (which an empty branch
  doesn't allow ;).

- you're calling get_export_generator without dest nor root and
  the code in bzrlib defaults to dest to set root. This raises an
  interesing point: which root should be used here (i.e. what do
  we want to prefix all the paths in the archive
  with). <project>-<branch nick>-<revno> may be nice (but ask
  others for feedback too).

- you used '.tar.gz' for the format but bzr expects either 'tgz'
  OR a dest file name to deduce the format from the file
  suffix. I just used 'tgz' there.

With these fixes in place we get:

======================================================================
ERROR: bzrlib.plugins.loggerhead.loggerhead.tests.test_controllers.TestDownloadTarballUI.test_download_tarball
----------------------------------------------------------------------
_StringException: Text attachment: log
------------
0.622 creating repository in file:///tmp/testbzr-KY_qfE.tmp/bzrlib.plugins.loggerhead.loggerhead.tests.test_controllers.TestDownloadTarballUI.test_download_tarball/work/.bzr/.
0.624 creating branch <bzrlib.branch.BzrBranchFormat7 object at 0x22ae990> in file:///tmp/testbzr-KY_qfE.tmp/bzrlib.plugins.loggerhead.loggerhead.tests.test_controllers.TestDownloadTarballUI.test_download_tarball/work/
0.631 trying to create missing lock '/tmp/testbzr-KY_qfE.tmp/bzrlib.plugins.loggerhead.loggerhead.tests.test_controllers.TestDownloadTarballUI.test_download_tarball/work/.bzr/checkout/dirstate'
0.631 opening working tree '/tmp/testbzr-KY_qfE.tmp/bzrlib.plugins.loggerhead.loggerhead.tests.test_controllers.TestDownloadTarballUI.test_download_tarball/work'
0.642 export version <InventoryRevisionTree instance at 29ece90, rev_id='null:'>
0.649 opening working tree '/tmp/testbzr-KY_qfE.tmp'
------------
Text attachment: traceback
------------
Traceback (most recent call last):
  File "/usr/lib/python2.7/dist-packages/testtools/runtest.py", line 169, in _run_user
    return fn(*args, **kwargs)
  File "/usr/lib/python2.7/dist-packages/testtools/testcase.py", line 499, in _run_test_method
    return self._get_test_method()()
  File "/home/vila/.bazaar/plugins/loggerhead/loggerhead/tests/test_controllers.py", line 140, in test_download_tarball
    res = app.get('/tarball')
  File "/usr/lib/pymodules/python2.7/paste/fixture.py", line 208, in get
    return self.do_request(req, status=status)
  File "/usr/lib/pymodules/python2.7/paste/fixture.py", line 389, in do_request
    **req.environ)
  File "/usr/lib/pymodules/python2.7/paste/wsgilib.py", li...

review: Needs Fixing

[Gavin Panella (allenap)]

Hi,

Please can you resolve the conflicts?

  Text conflict in loggerhead/apps/branch.py
  Text conflict in loggerhead/controllers/revision_ui.py
  Text conflict in loggerhead/tests/test_controllers.py

Gavin.

[xaav (xaav)]

Okay.

[lp:~xaav/loggerhead/export-tarball]

451. By xaav <email address hidden> on 2011-07-07

Merged lp:loggerhead

452. By xaav <email address hidden> on 2011-07-07

Fixed extension issue

[xaav (xaav)]

Is there any reason this is being held up? It worked fine for me.

[Martin Pool (mbp)]

[fix] It looks like you want tarballs disabled by default (which I agree with), so it seems a bit strange to me that the keyword arguments default to true.

[fix] You seem to be giving people uncompressed tarballs? .tgz is probably better. The default value to export_archive is unnecessary and probably just a distraction if it's called from only one place that does specify the value.

Were you able to get the tests to run and pass?

Aside from that it looks good to me.

review: Needs Fixing

[xaav (xaav)]

> [fix] You seem to be giving people uncompressed tarballs? .tgz is probably better. The default value to export_archive is unnecessary and probably just a distraction if it's called from only one place that does specify the value.

.tgz tarballs are not possible due to a bug in bzr. It is not within the scope of this patch to fix that bug.

> Were you able to get the tests to run and pass?

Yes. Everything passes.

> [fix] It looks like you want tarballs disabled by default (which I agree with), so it seems a bit strange to me that the keyword arguments default to true.

What line are you referring to?

[lp:~xaav/loggerhead/export-tarball]

453. By xaav on 2011-07-10

Fixed buggy merging.

454. By xaav <email address hidden> on 2011-07-10

Fixed buggy merging and removed IDE files

[Martin Pool (mbp)]

> .tgz tarballs are not possible due to a bug in bzr. It is not within the scope of this patch to fix that bug.

So, when we fix that bug, we'll change this to send tgz instead?

Could you tell me the bug number? Perhaps it would be worth
mentioning this in a comment
>
>
>> Were you able to get the tests to run and pass?
>
> Yes. Everything passes.

Great!

>
>> [fix] It looks like you want tarballs disabled by default (which I agree with), so it seems a bit strange to me that the keyword arguments default to true.
>
> What line are you referring to?
>
@@ -49,7 +49,7 @@

     def __init__(self, branch, friendly_name=None, config={},
                  graph_cache=None, branch_link=None, is_root=False,
- served_url=_DEFAULT, use_cdn=False):
+ served_url=_DEFAULT, use_cdn=False, export_tarballs=True):
         self.branch = branch
         self._config = config
         self.friendly_name = friendly_name

[xaav (xaav)]

> Could you tell me the bug number? Perhaps it would be worth mentioning this in a comment.

I really don't know if the bug has been reported. All I know is if it is changed to .tar.gz, a 500 internal server error occurs and the traceback goes right into bzrlib.

I'm not the only one with the problem, vila encountered the same problem. For more information, see his comment.

[lp:~xaav/loggerhead/export-tarball]

455. By xaav on 2011-07-11

Fixed serve tarballs issue.

[xaav (xaav)]

>
>> [fix] It looks like you want tarballs disabled by default (which I agree with), so it seems a bit strange to me that the keyword arguments default to true.
>
> What line are you referring to?
>
@@ -49,7 +49,7 @@

     def __init__(self, branch, friendly_name=None, config={},
                  graph_cache=None, branch_link=None, is_root=False,
- served_url=_DEFAULT, use_cdn=False):
+ served_url=_DEFAULT, use_cdn=False, export_tarballs=True):
         self.branch = branch
         self._config = config
         self.friendly_name = friendly_name

Fixed.

[Martin Pool (mbp)]

Oh I see. That looks pretty trivial; I'll see about fixing it.

[Martin Pool (mbp)]

[fix] You'll need to also remove this line:

  from bzrlib.export.tar_exporter import export_tarball

from history.py since it's not used and doesn't exist.

I fixed the unbound local warning here: https://code.launchpad.net/~mbp/bzr/trivial/+merge/67474

When I run the tests for this (with those two fixes) I get a failure in your new test_download_tarball, because the response is

303 See Other
The resource has been moved to /;
you should be redirected automatically.

which makes it look like it's not actually matching the URL. Do you see this too?

[Martin Pool (mbp)]

I'd really like to see this landed and live, so if you want any help just ask, here or on #bzr irc.

[xaav (xaav)]

> When I run the tests for this (with those two fixes) I get a failure in your new test_download_tarball, because the response is

This is probably due to revision 455. I'll see what I can do.

[xaav (xaav)]

Try now.

[lp:~xaav/loggerhead/export-tarball]

456. By xaav on 2011-07-11

Fixed tests.

[Martin Pool (mbp)]

I get the same failure. You don't?

[Martin Pool (mbp)]

However, changing it to just `res.body` fixes it, and then everything passes. I'll also add some checks that the response has the right status code and headers, and I'll see if we can change it to get a tgz now my fix has landed.

[xaav (xaav)]

> and I'll see if we can change it to get a tgz now my fix has landed.

This is bleeding-edge code, so you'll need to get lp to upgrade their bzr version.

[Martin Pool (mbp)]

On 11 July 2011 13:34, Xaav <email address hidden> wrote:
>> and I'll see if we can change it to get a tgz now my fix has landed.
>
> This is bleeding-edge code, so you'll need to get lp to upgrade their bzr version.

I think this is ok; landing a new bazaar at the same time as a new bzr
is no harder.

I'm going to switch it to sending tgz, and test it a bit
interactively, and I think then we'll be good to go.

[lp:~xaav/loggerhead/export-tarball]

457. By xaav on 2011-07-11

Added to UI.

[xaav (xaav)]

I reverted the change in 455, because I couldn't download tarballs via a web browser.

[Martin Pool (mbp)]

> I reverted the change in 455, because I couldn't download tarballs via a web
> browser.

Right, specifically turning it back on by default. I think that's fine.

I'm just testing it interactively in a browser and it does seem to work ok, which is great.

[fix] There is one bug which is that the tarball link always gets you the tarball from the current version of the branch, even if you're looking at an older version, which I think people would find confusing. Could you try to fix that?

My small changes are in lp:~mbp/loggerhead/export-tarball

[xaav (xaav)]

BTW, that link is broken.

[xaav (xaav)]

> [fix] There is one bug which is that the tarball link always gets you the tarball from the current version of the branch, even if you're looking at an older version, which I think people would find confusing. Could you try to fix that?

This is really a UI problem. Following the link /tarball/12 downloads revision 12, but I can't access the revision number from menu.pt, and I'm not that familiar w/ TAL.

[lp:~xaav/loggerhead/export-tarball]

458. By xaav on 2011-07-12

UI fixes.

[Martin Pool (mbp)]

Sorry for the delay - that looks good to me.

[Martin Pool (mbp)]

review: Approve

[Ian Booth (wallyworld)]

Given the number of eyeballs that have already seen this mp, it looks ok to me apart from some small lint issues. I can fix those and then land this mp.

review: Approve (code*)

[xaav (xaav)]

Is this merged? If so, should the bug be closed?

[Martin Pool (mbp)]

Hi xaav, it's not merged yet.

I spoke to Ian, and our plan is:
 * we will get a final review, maybe from John
 * Ian will test it out locally and if that's ok he will merge it in
to loggerhead trunk and shepherd deployment on lp

[John A Meinel (jameinel)]

[info] You are already exposing information "can_export" into the menu.pt. Why not just expose the revision number there instead. So "exported_tarball_revno" can be None indicating it has no tarball to download, or can be a string indicating what URL (or part of url, or whatever) to use.

[needsinfo] This defaults to always offering download links. Which means that if this rolls out on Launchpad, we will be letting people download via tarball. It hasn't been clear to me if this was specifically approved for Launchpad. As such *I* would tend to change:

25 def __init__(self, branch, friendly_name=None, config={},
26 graph_cache=None, branch_link=None, is_root=False,
27 - served_url=_DEFAULT, use_cdn=False):
28 + served_url=_DEFAULT, use_cdn=False, export_tarballs=True):

To "export_tarballs=False". Then if Launchpad wants to expose tarballs, it can chose to do so. We can easily update the 'bzr serve --http' and './serve_branches' code so that they default to exposing tarballs, without changing Launchpad's policy without their explicit review of it.

[needsinfo] This does all the exporting as raw '.tar' files. Without a way to configure it. One idea would be to change the 'export_tarballs' flag. Instead of being a boolean, it could be "tarball_format=None". If the format is None, then we don't export tarballs. Otherwise it can be 'tar' or 'tar.gz', or whatever.

I'm certainly open to discussion on these points, but I do think we should have answers for them before we land this.

review: Needs Information

[Gavin Panella (allenap)]

review: Abstain

[Robert Collins (lifeless)]

FWIW the default being on is fine - we can override when we upgrade
LP's revno. The format should default to something sensible like
tar.gz I think.

[Martin Pool (mbp)]

So the main remaining point then seems to be it might be nice to
compress them by default. But even there, it's pretty marginal
benefit.

And then it just needs someone to shepherd this through landing and deployment.

Anything else?

Martin

[Robert Collins (lifeless)]

Nothing else from me.

[Martin Pool (mbp)]

I will see about turning on compression, testing it, and landing it.

=== modified file '.bzrignore'
--- .bzrignore	2011-03-10 14:22:12 +0000
+++ .bzrignore	2011-07-12 03:33:24 +0000
@@ -8,3 +8,5 @@
 loggerhead-memprofile
 ./docs/_build/
 tags
+.project
+.pydevproject
 
=== modified file 'loggerhead/apps/branch.py'
--- loggerhead/apps/branch.py	2011-06-28 16:09:37 +0000
+++ loggerhead/apps/branch.py	2011-07-12 03:33:24 +0000
@@ -33,7 +33,7 @@
 from loggerhead.controllers.atom_ui import AtomUI
 from loggerhead.controllers.changelog_ui import ChangeLogUI
 from loggerhead.controllers.diff_ui import DiffUI
-from loggerhead.controllers.download_ui import DownloadUI
+from loggerhead.controllers.download_ui import DownloadUI, DownloadTarballUI
 from loggerhead.controllers.filediff_ui import FileDiffUI
 from loggerhead.controllers.inventory_ui import InventoryUI
 from loggerhead.controllers.revision_ui import RevisionUI
@@ -49,7 +49,7 @@
 
     def __init__(self, branch, friendly_name=None, config={},
                  graph_cache=None, branch_link=None, is_root=False,
-                 served_url=_DEFAULT, use_cdn=False):
+                 served_url=_DEFAULT, use_cdn=False, export_tarballs=True):
         self.branch = branch
         self._config = config
         self.friendly_name = friendly_name
@@ -61,6 +61,7 @@
         self.is_root = is_root
         self.served_url = served_url
         self.use_cdn = use_cdn
+        self.export_tarballs = export_tarballs
 
     def get_history(self):
         file_cache = None
@@ -126,6 +127,7 @@
         'revision': RevisionUI,
         'search': SearchUI,
         'view': ViewUI,
+        'tarball': DownloadTarballUI,
         }
 
     def last_updated(self):
 
=== modified file 'loggerhead/config.py'
--- loggerhead/config.py	2010-05-12 14:38:05 +0000
+++ loggerhead/config.py	2011-07-12 03:33:24 +0000
@@ -36,6 +36,7 @@
         use_cdn=False,
         sql_dir=None,
         allow_writes=False,
+        export_tarballs=True,
         )
     parser.add_option("--user-dirs", action="store_true",
                       help="Serve user directories as ~user.")
@@ -75,6 +76,8 @@
                       help="The directory to place the SQL cache in")
     parser.add_option("--allow-writes", action="store_true",
                       help="Allow writing to the Bazaar server.")
+    parser.add_option("--export-tarballs", action="store_true",
+                      help="Allow exporting revisions to tarballs.")
     return parser
 
 
 
=== modified file 'loggerhead/controllers/__init__.py'
--- loggerhead/controllers/__init__.py	2011-06-28 16:09:37 +0000
+++ loggerhead/controllers/__init__.py	2011-07-12 03:33:24 +0000
@@ -21,7 +21,7 @@
 import simplejson
 import time
 
-from paste.httpexceptions import HTTPNotFound
+from paste.httpexceptions import HTTPNotFound, HTTPSeeOther
 from paste.request import path_info_pop, parse_querystring
 
 from loggerhead import util
 
=== modified file 'loggerhead/controllers/download_ui.py'
--- loggerhead/controllers/download_ui.py	2010-05-05 19:03:40 +0000
+++ loggerhead/controllers/download_ui.py	2011-07-12 03:33:24 +0000
@@ -19,46 +19,51 @@
 
 import logging
 import mimetypes
+import os
 import urllib
 
 from paste import httpexceptions
 from paste.request import path_info_pop
 
 from loggerhead.controllers import TemplatedBranchView
+from loggerhead.exporter import export_archive
 
 log = logging.getLogger("loggerhead.controllers")
 
 
 class DownloadUI (TemplatedBranchView):
 
-    def __call__(self, environ, start_response):
-        # /download/<rev_id>/<file_id>/[filename]
-
-        h = self._history
-
+    def encode_filename(self, filename):
+
+        return urllib.quote(filename.encode('utf-8'))
+
+    def get_args(self, environ):
         args = []
         while True:
             arg = path_info_pop(environ)
             if arg is None:
                 break
             args.append(arg)
+        return args
 
+    def __call__(self, environ, start_response):
+        # /download/<rev_id>/<file_id>/[filename]
+        h = self._history
+        args = self.get_args(environ)
         if len(args) < 2:
             raise httpexceptions.HTTPMovedPermanently(
                 self._branch.absolute_url('/changes'))
-
         revid = h.fix_revid(args[0])
         file_id = args[1]
         path, filename, content = h.get_file(file_id, revid)
         mime_type, encoding = mimetypes.guess_type(filename)
         if mime_type is None:
             mime_type = 'application/octet-stream'
-
         self.log.info('/download %s @ %s (%d bytes)',
                       path,
                       h.get_revno(revid),
                       len(content))
-        encoded_filename = urllib.quote(filename.encode('utf-8'))
+        encoded_filename = self.encode_filename(filename)
         headers = [
             ('Content-Type', mime_type),
             ('Content-Length', str(len(content))),
@@ -67,3 +72,29 @@
             ]
         start_response('200 OK', headers)
         return [content]
+
+
+class DownloadTarballUI(DownloadUI):
+
+    def __call__(self, environ, start_response):
+        """Stream a tarball from a bazaar branch."""
+        # Tried to re-use code from downloadui, not very successful
+        format = "tar"
+        history = self._history
+        self.args = self.get_args(environ)
+        if len(self.args):
+            revid = history.fix_revid(self.args[0])
+        else:
+            revid = self.get_revid()
+        if self._branch.export_tarballs:
+            root = 'branch'
+            encoded_filename = self.encode_filename(root + '.' + format)
+            headers = [
+                ('Content-Type', 'application/octet-stream'),
+                ('Content-Disposition',
+                 "attachment; filename*=utf-8''%s" % (encoded_filename,)),
+                ]
+            start_response('200 OK', headers)
+            return export_archive(history, root, revid, format)
+        else:
+            raise httpexceptions.HTTPSeeOther('/')
 
=== modified file 'loggerhead/controllers/revision_ui.py'
--- loggerhead/controllers/revision_ui.py	2011-06-27 17:11:24 +0000
+++ loggerhead/controllers/revision_ui.py	2011-07-12 03:33:24 +0000
@@ -136,6 +136,7 @@
                 self._branch.friendly_name,
                 self._branch.is_root,
                 'changes'))
+        can_export = self._branch.export_tarballs
 
         values.update({
             'history': self._history,
@@ -149,6 +150,7 @@
             'filter_file_id': filter_file_id,
             'diff_chunks': diff_chunks,
             'query': query,
+            'can_export': can_export,
             'specific_path': path,
             'start_revid': start_revid,
         })
 
=== added file 'loggerhead/exporter.py'
--- loggerhead/exporter.py	1970-01-01 00:00:00 +0000
+++ loggerhead/exporter.py	2011-07-12 03:33:24 +0000
@@ -0,0 +1,46 @@
+# Copyright (C) 2011 Canonical Ltd
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+"""Exports an archive from a bazaar branch"""
+
+from bzrlib.export import get_export_generator
+
+
+class ExporterFileObject(object):
+
+    def __init__(self):
+        self._buffer = []
+
+    def write(self, s):
+        self._buffer.append(s)
+
+    def get_buffer(self):
+        try:
+            return ''.join(self._buffer)
+        finally:
+            self._buffer = []
+
+
+def export_archive(history, root, revid, format=".tar.gz"):
+    """Export tree contents to an archive
+
+    :param history: Instance of history to export
+    :param root: Root location inside the archive.
+    :param revid: Revision to export
+    :param format: Format of the archive
+    """
+    fileobj = ExporterFileObject()
+    tree = history._branch.repository.revision_tree(revid)
+    for _ in get_export_generator(tree=tree, root=root, fileobj=fileobj, format=format):
+        yield fileobj.get_buffer()
+    # Might have additonal contents written
+    yield fileobj.get_buffer()
 
=== modified file 'loggerhead/history.py'
--- loggerhead/history.py	2011-03-25 13:09:10 +0000
+++ loggerhead/history.py	2011-07-12 03:33:24 +0000
@@ -33,6 +33,7 @@
 import re
 import textwrap
 import threading
+import tarfile
 
 from bzrlib import tag
 import bzrlib.branch
@@ -44,6 +45,7 @@
 from loggerhead import search
 from loggerhead import util
 from loggerhead.wholehistory import compute_whole_history_data
+from bzrlib.export.tar_exporter import export_tarball
 
 
 def is_branch(folder):
@@ -816,4 +818,6 @@
             renamed=sorted(reporter.renamed, key=lambda x: x.new_filename),
             removed=sorted(reporter.removed, key=lambda x: x.filename),
             modified=sorted(reporter.modified, key=lambda x: x.filename),
-            text_changes=sorted(reporter.text_changes, key=lambda x: x.filename))
+            text_changes=sorted(reporter.text_changes,
+                                key=lambda x: x.filename))
+
 
=== added directory 'loggerhead/static/downloads'
=== modified file 'loggerhead/templates/menu.pt'
--- loggerhead/templates/menu.pt	2011-02-09 22:33:59 +0000
+++ loggerhead/templates/menu.pt	2011-07-12 03:33:24 +0000
@@ -6,6 +6,8 @@
                                id string:on">Changes</a></li>
         <li><a tal:attributes="href python:url('/files', clear=1);
                                title string:Files">Files</a></li>
+        <li><a tal:attributes="href python:url(['/revision', change.revno], clear=1);
+                               title string:Tarball">Tarball</a></li>
         </tal:changes-active>
         <tal:fileview-active tal:condition="fileview_active">
         <li><a tal:attributes="href python:url('/changes', clear=1);
@@ -13,6 +15,8 @@
         <li><a tal:attributes="href python:url('/files', clear=1);
                                title string:Files;
                                id string:on">Files</a></li>
+        <li><a tal:attributes="href python:url(['/revision', change.revno], clear=1);
+                               title string:Tarball">Tarball</a></li>
         </tal:fileview-active>
       </ul>
 </tal:menu>
 
=== modified file 'loggerhead/templates/revision.pt'
--- loggerhead/templates/revision.pt	2011-06-28 16:51:55 +0000
+++ loggerhead/templates/revision.pt	2011-07-12 03:33:24 +0000
@@ -84,7 +84,10 @@
                tal:attributes="href python:url(['/diff', change.revno], clear=1)">download diff</a>
             <a tal:condition="python:compare_revid is not None"
                tal:attributes="href python:url(['/diff', change.revno, history.get_revno(compare_revid)], clear=1)">download diff</a>
-          </li>
+           </li>
+           <li tal:condition="python:can_export">
+            <a tal:attributes="href python:url(['/tarball', change.revno])">Download tarball</a>
+           </li>
           <li id="last"><a tal:attributes="href python:url(['/changes', change.revno]);
                                            title string:view history from revision ${change/revno}"
                            tal:content="string:view history from revision ${change/revno}"></a></li>
 
=== modified file 'loggerhead/tests/test_controllers.py'
--- loggerhead/tests/test_controllers.py	2011-06-28 16:06:12 +0000
+++ loggerhead/tests/test_controllers.py	2011-07-12 03:33:24 +0000
@@ -1,3 +1,10 @@
+from cStringIO import StringIO
+import logging
+import tarfile
+
+from paste.httpexceptions import HTTPServerError
+
+from bzrlib import errors
 import simplejson
 
 from loggerhead.apps.branch import BranchWSGIApp
@@ -204,3 +211,16 @@
         revlog_ui = branch_app.lookup_app(env)
         self.assertOkJsonResponse(revlog_ui, env)
 
+class TestDownloadTarballUI(BasicTests):
+
+    def setUp(self):
+        super(TestDownloadTarballUI, self).setUp()
+        self.createBranch()
+
+    def test_download_tarball(self):
+        app = self.setUpLoggerhead()
+        res = app.get('/tarball')
+        f = open('tarball', 'w')
+        f.write(res)
+        f.close()
+        self.failIf(not tarfile.is_tarfile('tarball'))
 \ No newline at end of file
 
=== modified file 'loggerhead/tests/test_simple.py'
--- loggerhead/tests/test_simple.py	2011-06-28 14:19:58 +0000
+++ loggerhead/tests/test_simple.py	2011-07-12 03:33:24 +0000
@@ -46,7 +46,7 @@
         self.tree = self.make_branch_and_tree('.')
 
     def setUpLoggerhead(self, **kw):
-        branch_app = BranchWSGIApp(self.tree.branch, '', **kw).app
+        branch_app = BranchWSGIApp(self.tree.branch, '', export_tarballs=True, **kw).app
         return TestApp(HTTPExceptionHandler(branch_app))
 
     def assertOkJsonResponse(self, app, env):