Missing separator on macro call ?

[Magnus Ihse Bursie]

On 2017-05-29 12:25, Kim Barrett wrote:
>> On May 29, 2017, at 3:52 AM, David Holmes <david.holmes at oracle.com> wrote:
>>
>> On 29/05/2017 5:25 PM, Magnus Ihse Bursie wrote:
>>> Maybe we should have a way to scan the source tree for files with problematic names, like *#* or *,*...? Or CachedFind should be written to handle *#* files properly?
>> I have a recollection that these temp files used to cause a different kind of error that was much more obvious.
> I have the same recollection.

I think I remember now. The failure discussed then was for a file in the 
hotspot directory. When we wrapped the old hotspot build, we first 
scanned the entire hotspot folder to see if anything had changed, since 
the "nop" build was so slow with the old hotspot build. This failed, but 
with a more obvious error result.

This scanning is no longer present. Also, depending on where in the 
source tree a spurious file may appear, different pieces of make logic 
might run into it first, so it's impossible to say how things will break.

/Magnus

>
>> I remember commenting that it was good that the build failed because you wouldn't be building waht you expected. But this "missing separator" error is just ridiculously obscure. :( Though something you should only get bitten by once.
> I ran into it just recently.  Thankfully, I remembered this thread...
>