py3: fix curses chunkselector fallback (when diffs are too large) on py3 Previously we showed the message using Exception.message, which is removed in py3. Since crecordmod.fallbackerror inherits from error.Abort, we can just use `b'%s' % exception` to print the message. This does not print the hint, but that's fine - we don't set one. We inherit from error.Abort so that if a codepath doesn't handle fallback specially, it exits to the terminal with a sane message instead of an unknown exception error. Differential Revision: https://phab.mercurial-scm.org/D7912

transaction: allow finalizer to add finalizer It will make some code (persistent nodemap related) simpler to write, because higher level code can blindly queue finalization without thinking too hard about the context. Differential Revision: https://phab.mercurial-scm.org/D7833

tests: stabilize test-subrepo-svn.t on Windows This partially reverts 991e4404e910, because the URL form of `C:\...` gets escaped to `C%3A/...`, which breaks the substitution of $TESTTMP. The forget command on 'notafile*' errored out with: notafile*: The filename, directory name, or volume label syntax is incorrect which I think is because '*' isn't a legal character in a file name (though I can't trigger this directly from MSYS or cmd.exe for some reason). Differential Revision: https://phab.mercurial-scm.org/D7816

rebase: fix bug where `--collapse` would apply diff on missing file Even though the file was missing, the rebase would succeed. Differential Revision: https://phab.mercurial-scm.org/D7897

rebase: extract a variable for a repeated `repo[p1]` I'll add another use site in the next patch. Differential Revision: https://phab.mercurial-scm.org/D7896

graftcopies: document why the function is useful at all Despite having spent a significant amount on time on the copy-tracing code, I thought `graftcopies()` (formerly known as `duplicatecopies()`) was needed to duplicate copies after calling `merge.update()` to do a merge (as `merge.graft()` does), but it's actually usually not needed; `merge.update()` takes care of most copies. This patch documents what the function is for. Differential Revision: https://phab.mercurial-scm.org/D7861

graftcopies: remove `skip` and `repo` arguments The `skip` argument was added in 2ba6c9b4e0eb (rebase: fix bug that caused transitive copy records to disappear (issue4192), 2014-06-07) in order to fix https://bz.mercurial-scm.org/show_bug.cgi?id=4192. I ran tests at that commit without the `skiprev` argument and the only difference I noticed was that `test-rebase-collapse.t` failed differently, in the call that is now on line 501. Without the `skiprev` argument, that call would end up creating another commit because it tried to record an invalid copy. With the previous patch in this series, such invalid copies are no longer recorded, so it seems we don't need the `skip` argument anymore. I also removed the `repo` argument since that also becomes unused with the removal of the `skip` argument. Differential Revision: https://phab.mercurial-scm.org/D7860

graftcopies: use _filter() for filtering out invalid copies `graftcopies()` (formerly called `duplicatecopies()`) checked that the copy destination existed in the working copy, but it didn't check that copy source existed in the parent of the working copy. In `test-graft.t` we can see that as warnings about not finding ancestors of the copied files, and also empty commits getting created. This patch uses the existing `_filter()` function for filtering out invalid copies. In addition to the aforementioned types, that also includes copies where source and destination is the same. Differential Revision: https://phab.mercurial-scm.org/D7859

copies: replace duplicatecopies() by function that takes contexts The callers mostly have context objects, so let's avoid looking up the same context objects inside `duplicatecopies()`. I also renamed the function to `graftcopies()` since I think that better matches its purpose. I did it in the same commit so it's easier for extensions to switch between the functions. Differential Revision: https://phab.mercurial-scm.org/D7858

graft: extract repo[None] to a variable I plan to allow the caller pass in an overlayworkingctx, so this prepares for that. Differential Revision: https://phab.mercurial-scm.org/D7857

rust-core: fix typo in comment Differential Revision: https://phab.mercurial-scm.org/D7895

sha1dc: use buffer protocol when parsing arguments Without this, functions won't accept bytearray, memoryview, or other types that can be exposed as bytes to the C API. The most resilient way to obtain a bytes-like object from the C API is using the Py_buffer interface. This commit converts use of s#/y# to s*/y* and uses Py_buffer for accessing the underlying bytes array. I checked how hashlib is implemented in CPython and the the implementation agrees with its use of the Py_buffer interface as well as using BufferError in cases of bad buffer types. Sadly, there's no good way to test for ndim > 1 without writing our own C-backed Python type. Differential Revision: https://phab.mercurial-scm.org/D7879

lfs: avoid quadratic performance in processing server responses This is also adapted from the Facebook repo[1]. Unlike there, we were already reading the download stream in chunks and immediately writing it to disk, so we basically avoided the problem on download. There shouldn't be a lot of data to read on upload, but it's better to get rid of this pattern. [1] https://github.com/facebookexperimental/eden/commit/82df66ffe97e21f3ee73dfec093c87500fc1f6a7 Differential Revision: https://phab.mercurial-scm.org/D7882

lfs: check content length after downloading content Adapted from the Facebook repo[1]. The intent is to distinguish between the connection dying and getting served a corrupt blob. The original message: HTTP makes no provision to tell your client that you failed halfway through producing your response and won't have the answer they're looking for. So, if a LFS server fails while producing a response, then we'll report an OID mismatch. We can do a little better and disambiguate between "the server sent us the wrong blob" (very scary) and "the server crashed" (merely annoying) by looking at the content length of the response we got back. If it's not what was advertised, we can reasonably safely assume the server crashed. [1] https://github.com/facebookexperimental/eden/commit/2a4a6fab4e882ed89b948bfc1e7d56d7c3c99dd2 Differential Revision: https://phab.mercurial-scm.org/D7881

lfs: rename a variable to clarify its use This is the response object, not a request. Differential Revision: https://phab.mercurial-scm.org/D7880

sha1dc: use proper string functions on Python 2/3 PyString_FromStringAndSize doesn't exist on Python 3: we need to use PyUnicode_FromStringAndSize. The extension now compiles without warnings on Python 2 and 3. Differential Revision: https://phab.mercurial-scm.org/D7878