At 8:44 AM +0200 on 8/22/10, Martin =?UTF-8?B?VHLDvGJuZXI=?= wrote about Re: LARL vs. Literal Alignment:
But it lacks the beauty of a solution invisable to the HLASM-coder. The "invisable" solution would generate one literal for these two (I would expect): LARL R1,=C'THIS IS A SAMPLE' LA R1,=C'THIS IS A SAMPLE'
As well as a sanity check that would force the correct alignment to the literal even if the first occurrence was the LA one (IOW: When the LARL version is encountered generate it and alter the LA's reference to use it).