[LLVMdev] test suite 'owner'
echristo at gmail.com
Mon Jan 13 14:36:23 PST 2014
On Mon, Jan 13, 2014 at 12:36 PM, Robert Lytton <robert at xmos.com> wrote:
> ... and so (I infer from that) it should not be patched let alone need any
There are the occasional system level changes to it, but they're rare.
> Assuming my inference is correct, any patching should only affect the XCore
> target and only if there is a good reason why the XCore requires the change.
> So, is #ifdef around all/most changes the correct way to submit a patch?
It's painful, but if there's no other way...
> From: Eric Christopher [echristo at gmail.com]
> Sent: 13 January 2014 20:24
> To: Robert Lytton; llvmdev at cs.uiuc.edu
> Subject: RE: [LLVMdev] test suite 'owner'
> The idea is that it's going to be a correct and portable set of code that
> works both as a correctness and performance test suite.
> On Mon Jan 13 2014 at 12:22:51 PM, Robert Lytton <robert at xmos.com> wrote:
>> Hi Eric,
>> Could you explain the intent and policy regarding the test-suite body of
>> Should the test be left as much as possible as-is (even if technically
>> Should changes only affect the XCore target (#ifdef) or should all targets
>> get the changes?
>> Taking "int32_t main" as an example.
>> The correct return type & argc for main is 'int'.
>> In the XCore tool chain, 'int32_t' equates to long (IIRC) and hence is not
>> acceptable in the type signature for main.
>> Should this change be only for the XCore target or all targets?
>> When I know the policy for the test-suite, I'll alter as necessary &
>> regroup the changes into patches containing the same type of change and
>> submit for approval.
>> One more question:
>> On patch I need to address is how to make deterministic the order of
>> stdout & stderr.
>> Ideally, applications would use either stdout or stderr but not both.
>> Would a patch to change to only stdout be acceptable (plus any changes to
>> expected output)?
>> Thank you
>> From: Eric Christopher [echristo at gmail.com]
>> Sent: 13 January 2014 19:16
>> To: Robert Lytton; llvmdev at cs.uiuc.edu
>> Subject: [LLVMdev] test suite 'owner'
>> Some of these are pretty weird, e.g. int32_t main. Probably the best thing
>> is to submit each patch individually with an explanation of what the purpose
>> is and we can talk about them then.
>> On Fri Jan 10 2014 at 4:13:47 AM, Robert Lytton <robert at xmos.com> wrote:
>> I have found it necessary to make some changes to the test-suite for the
>> XCore platform.
>> These changes include:
>> altering #includes, as supported by XCore;
>> using stdout or stderr to make the output diffs consistent (fixing
>> expected output too);
>> (This work is still under review as best way to do it)
>> 'fixing' symbol and type problems e.g name clashes & scope;
>> altering/adding code snippets and macros.
>> I have used #ifdef to limit and keep any changes specific to the XCore.
>> Some of these could/should be made common to all targets e.g. log2() ->
>> I have also altered the Makefile to filter out tests not supported by the
>> I would like to discuss the changes I have found necessary to make and
>> what is the next step.
>> Should any/all of them be pushed upstream?
More information about the llvm-dev