• "Matthew Wilson" <matthew thedjournal.com> Jun 13 2002
• "Walter" <walter digitalmars.com> Jun 14 2002
• "Matthew Wilson" <dmd synesis.com.au> Jun 14 2002
• "Walter" <walter digitalmars.com> Jun 14 2002
• "Matthew Wilson" <dmd synesis.com.au> Jun 15 2002

Walter

Am compiling some moderately complex template stuff, and with v8.27 I get

    Internal error: init 1814

whereas with v8.28 I get

    Internal error: init 1815

Obviously if I want you to be able to fix it I'll have to boil down the code
(there's 900 lines of cpp, and about 2500 lines of header), but I would like
to know first whether the increment in internal error is a function of a
similar-sized (ie. by 1) increment in the compiler version number, or
whether it is likely to be an artefact of the later compiler getting a
little further before apoptosis?

Thanks

Matthew

"Matthew Wilson" <matthew thedjournal.com> wrote in message
news:aec14p$22pt$1 digitaldaemon.com...
 Walter

 Am compiling some moderately complex template stuff, and with v8.27 I get

     Internal error: init 1814

 whereas with v8.28 I get

     Internal error: init 1815

 Obviously if I want you to be able to fix it I'll have to boil down the


 (there's 900 lines of cpp, and about 2500 lines of header), but I would


 to know first whether the increment in internal error is a function of a
 similar-sized (ie. by 1) increment in the compiler version number, or
 whether it is likely to be an artefact of the later compiler getting a
 little further before apoptosis?


The number is simply the line number in the source code that threw the
assert. I would appreciate it if you could boil it down, in my experience
about 97% of all such bugs can be reduced to 10 lines of code or less. That
makes it much easier to fix. -Walter

The line number thing is a little strange because it's exactly the same code
that's being compiled, just with different versions. Does one do 0-based and
the other do 1-based? Surely not?

Just checked, and the source file has 847 lines. Only two headers are > 1000
lines, and only one has anything like code on or around lines 1814/5. Very
odd.

I'll try and boil it down.


"Walter" <walter digitalmars.com> wrote in message
news:aeda2b$c9i$1 digitaldaemon.com...
 "Matthew Wilson" <matthew thedjournal.com> wrote in message
 news:aec14p$22pt$1 digitaldaemon.com...
 Walter

 Am compiling some moderately complex template stuff, and with v8.27 I




     Internal error: init 1814

 whereas with v8.28 I get

     Internal error: init 1815

 Obviously if I want you to be able to fix it I'll have to boil down the


 (there's 900 lines of cpp, and about 2500 lines of header), but I would


 to know first whether the increment in internal error is a function of a
 similar-sized (ie. by 1) increment in the compiler version number, or
 whether it is likely to be an artefact of the later compiler getting a
 little further before apoptosis?


 The number is simply the line number in the source code that threw the
 assert. I would appreciate it if you could boil it down, in my experience
 about 97% of all such bugs can be reduced to 10 lines of code or less.


 makes it much easier to fix. -Walter

No, it's the line number in the compiler source.

"Matthew Wilson" <dmd synesis.com.au> wrote in message
news:aeelmm$1mek$1 digitaldaemon.com...
 The line number thing is a little strange because it's exactly the same


 that's being compiled, just with different versions. Does one do 0-based


 the other do 1-based? Surely not?

 Just checked, and the source file has 847 lines. Only two headers are >


 lines, and only one has anything like code on or around lines 1814/5. Very
 odd.

Duh, don't I feel silly

"Walter" <walter digitalmars.com> wrote in message
news:aeemek$1n6a$1 digitaldaemon.com...
 No, it's the line number in the compiler source.

 "Matthew Wilson" <dmd synesis.com.au> wrote in message
 news:aeelmm$1mek$1 digitaldaemon.com...
 The line number thing is a little strange because it's exactly the same


 that's being compiled, just with different versions. Does one do 0-based


 the other do 1-based? Surely not?

 Just checked, and the source file has 847 lines. Only two headers are >


 lines, and only one has anything like code on or around lines 1814/5.




 odd.