[lld] r193362 - Revert "r193300 - [PassManager] add ReaderWriter{Native, YAML} to the Driver"

[Rui Ueyama]

On Thu, Oct 24, 2013 at 11:12 PM, Shankar Easwaran
<shankare at codeaurora.org>wrote:

> Hi Rui, All,
>
> I mirror Chandlers thought that we really dont want to go in the path of
> adding these passes from each flavor, as thats only a change which would be
> removed by subsequent commits.
>
> Meanwhile, I have fixed all the tests across all flavors except the same
> failure across two different tests in pecoff.
>
> The test is drective.test/dynamic.test which is failing to create the
> entries in the Import table :-
>
> <------------snip-------------**------>
> The Import Tables:
>   lookup 00002068 time 00000000 fwd 00000000 name 000020a4 addr 00002088
>
>     DLL Name: vars.dll
>     Hint/Ord  Name
>            0  _name_with_underscore
>            1  fn
> <------------snip-------------**------>
>
> Can you take a look into that ? I have debugged that for a while and cant
> find a solution. On first thought I think there need to be seperate types
> for the derived atoms from the IDataAtom, and the Writer need to seperately
> handle writing of these atoms.
>
> We could need to disable 2 of these tests that fail due to the same issue,
> so that you can look into the failure, after the changes are committed
> which fixes all the other failures(Sent a review request --
> http://llvm-reviews.chandlerc.**com/D2022<http://llvm-reviews.chandlerc.com/D2022>
> )
>

Please do not submit the change that breaks tests. We need to keep the tree
always healthy. I will apply the patch locally and try to fix the test
failure myself, so please hold on.

Thanks
>
> Shankar Easwaran
>
>
> On 10/24/2013 5:21 PM, Rui Ueyama wrote:
>
>> By pushing I'm not referencing to this particular thing.
>>
>> Do you think you can enable the tests per-port basis, as Reid suggested? I
>> think that's better than guarding it with #ifdef, as you can enable it now
>> for ELF (so that it won't bit rotten while being disabled by #ifdef) and
>> other port owners can fix it independently.
>>
>>
>> On Thu, Oct 24, 2013 at 2:35 PM, Shankar Easwaran
>> <shankare at codeaurora.org>**wrote:
>>
>>  On 10/24/2013 4:27 PM, Rui Ueyama wrote:
>>>
>>>  LLD is still in an early stage of development. We've got many things to
>>>> do,
>>>> and this is not the highest priority among them that needs to be done by
>>>> next week or otherwise we should break Mach-O and COFF ports. I thank
>>>> you
>>>> for putting your effort on this new test scheme, but I can't promise we
>>>> can
>>>> fix them in such a short period of time.
>>>>
>>>> I hate to say but I'm honestly disappointed. We should respect each
>>>> other
>>>> in open source development, rather than forcing each other to do
>>>> something.
>>>> If you find someone have a different opinion, you need to elaborate why
>>>> you
>>>> think in a different way, rather than replying with a short email
>>>> repeating
>>>> basically the same thing. These days, virtually you and me submitted 90%
>>>> of
>>>> patches to LLD (I think I submitted like 60%), but I sometimes found
>>>> that
>>>> you pushed your opinion too hard without enough explanation. That's not
>>>> good. Please respect each other, or this development style does not
>>>> work.
>>>>
>>>>  I am so sorry If I implied that, I was only pushing it because this was
>>> one of the design goals (the Atom IR) of lld.
>>>
>>>
>>> Thanks
>>>
>>> Shankar Easwaran
>>>
>>> --
>>> Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted
>>> by the Linux Foundation
>>>
>>>
>>>
>
> --
> Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted
> by the Linux Foundation
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20131025/d13dcd29/attachment.html>