Re: [Chicken-hackers] ssql-record egg: ok -> fail on Salmonella report

[Peter Bex]

On Mon, Dec 25, 2017 at 11:11:34PM -0200, Arthur Maciel wrote:
> Dear Mario and friends, merry christmas and a happy new year!

Hi Arthur,

Best wishes to you and yours, as well!

> Does anyone have any idea why ssql-record egg started failing on salmonella
> exactly after April 18th 2017?
> 
> Ok:
> https://salmonella-linux-x86-64.call-cc.org/chicken-4/gcc/linux/x86-64/2017/04/18/salmonella-report/test/ssql-record.html
> 
> Fail:
> https://salmonella-linux-x86-64.call-cc.org/chicken-4/gcc/linux/x86-64/2017/04/19/salmonella-report/test/ssql-record.html
> 
> Am I missing some big change on CHICKEN's internals?

No, we just fixed some bugs, as you pointed out in your other two mails.

> I've been banging my head on the wall in the last 3 hours to make it work
> again, but I have no clue. Strangely some symbols that were expected to
> keep their 'names' intact are renamed internally (like "rec" -> "rec21")
> and this messes with the huge (and ugly) macro that I use.

Sorry to hear this has cost you so much time.  I had a quick look, and it's
pretty simple really.  It appears that you used the "inject" procedure to
undo syntax mangling from the input, which is an implementation detail of
how it worked in old versions of CHICKEN.

The "inject" procedure is intended to be used when you have constructed an
identifier and wish to protected it from being mangled in the macro's
output, instead.  To undo the name mangling of the input, you can use
strip-syntax.

Attached is a patch for your egg to fix this issue; with the patch, it
works in old and in new versions of CHICKEN.

Cheers,
Peter

0001-Fix-broken-ir-macro-injection.patch
Description: Text Data

signature.asc
Description: PGP signature