httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From dean gaudet <>
Subject Re: apr_sucks
Date Tue, 15 Aug 2000 12:24:59 GMT
On Mon, 14 Aug 2000, Rodent of Unusual Size wrote:

> dean gaudet wrote:
> > 
> > capricious symbol name changes are totally lame.
> We have code that's intended and designed to be reusable,
> and code that most definitely is neither.  Using the same
> namespace prefix for both is what's lame.

the example dougm gave which caused you all to jump in and make a symbol
name change is equivalent to someone saying "i want to link half of the
linux 2.2.17 kernel into my linux 2.0.36 kernel, do you think you could
accomodate by renaming everything in 2.2.17?"

if "half" isn't concrete enough for you then replace it with "the new
2.2.x dcache technology", or "the fine-threaded TCP stack", or any number
of new improvements in 2.2 that are fundamentally incompatible with 2.0.

you know guys, just because it's called a "module", doesn't mean it's
modular.  there's this continual trend of symbol name games... if it's not
because someone wants some name to be "pretty", then it's because someone
wants to link 5 billion libraries into the same executable.  this isn't
good software engineering.

i don't consider there to be any hard requirement that APR be linkable
with apache-1.3.

wasn't it wasteful enough doing the first s/pool/context/, then the
s/context/pool/, and now s/ap_pool_t/apr_pool_t/?  this is bloody

do you know how impossible it is to do comparisons across any amount of
time on our source trees any more?

do you realise that this means that you are compromising the security,
reliability, and robustness of the software?  how you ask?  well because
there are people out there who look at diffs and spot bugs.  but they
can't do that if the diffs are so fucking huge that they consist
essentially of the entire source tree.

symbol name changes (with no semantic change), formatting changes, these
are capricious, and detrimental.


View raw message