<div dir="ltr">On Mon, Oct 28, 2013 at 11:55 AM, Shankar Easwaran <span dir="ltr"><<a href="mailto:shankare@codeaurora.org" target="_blank">shankare@codeaurora.org</a>></span> wrote:<br><div class="gmail_extra"><div class="gmail_quote">
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><div class="im"><br>
On 10/28/2013 1:26 PM, Rui Ueyama wrote:<br>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">
Ah, there was a misunderstanding there. What I meant was that patch file<br>
attachments, as opposed to setting up a new git repository, is of course OK.<br>
</blockquote></div>
Ah.<div class="im"><br>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">
r193478 and r193477 are LGTM. They would fix the tests failing with<br>
RoundTrip tests. r193479 is however wouldn't.<br>
</blockquote></div>
These changes did fix some more of the failures.<br>
<br>
For example I couldnt figure out where the ImportDirectoryAtom gets added to the PassManager file.<br><br>
For example, the lld Pass calls createImportDirectory, which adds to the list of Import Directories.<br>
<br>
I looked at the YAML files too before adding those changes, and they didnt appear in the YAML file.<br></blockquote><div><br></div><div>All atoms whose are derived from IdataAtom are automatically added to context.file. Look at IdataAtom::IdataAtom(Context &, vector<uint_8>). It's weird if r193479 fixed the issue, but I might miss something.</div>
<div><br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">
I think you forgot to answer the question, are you ok with submitting the RoundTrip{YAML, Native} changes with the changes in the Driver removed, this would make all the tests pass and we can continue to figure out why those tests would fail, by enabling the RoundTrip code.<br>
</blockquote><div><br></div><div>In order to answer that question, I'd want to see the patch, so I asked for the latest patch. :)</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">
Once all the test results are fine, we could enable it completely.<br>
<br>
Looks like the IdataPass for COFF is really very complex!<br></blockquote><div><br></div><div>Yes, the pass creates a pretty complex data structure. It's complex partly because the file format is complex, but I'd suspect it's also because it's a pass. I have an idea to move some code from IdataPass to the writer to off-load some work that doesn't have to be done in a pass. I do not have time to try now, but refactoring is on my TODO list.</div>
</div></div></div>