TEAM-ADA Archives

Team Ada: Ada Programming Language Advocacy


Options: Use Classic View

Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Topic: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
"Robert C. Leif, Ph.D." <[log in to unmask]>
Mon, 28 Jun 1999 07:23:44 -0700
text/plain (46 lines)
From: Bob Leif
[log in to unmask]

To: W. Wesley Groleau and Team Ada

It takes longer to insert the e-mail address for this note using Microsoft
Outlook Express, than it does to compile a package body with Aonix Object
Ada under Windows 98. The slowest part of Object Ada is the Microsoft
linker. The problems with Ada compilation speed and CPU requirements are now
ancient history. Any computer than permits the reasonable use of Microsoft
Office is sufficient for Ada software development. In fact, the situation is
now reversed. The compiler manufactures can now afford to reduce the
compilation speed in the case of providing a significant benefit, such as
execution efficiency or improved error messages.

> -----Original Message-----
> From: Team Ada: Ada Advocacy Issues (83 & 95)
> [mailto:[log in to unmask]]On Behalf Of W. Wesley Groleau x4923
> Sent: Monday, June 28, 1999 6:51 AM
> To: [log in to unmask]
> Subject: Re: [5] Compilation speed data available?

> > On a mailing list concerned with another programming language (but not a
> > very well-known one*), the following comment appeared today:
> >
> > > If you want to see a really slow compiler I suggest
> > > you try using ADA (yuk!) if you haven't already.
> >
> > I suspect this may be based on not-very-current experience, and I would
> > like to provide some solid evidence that Ada compilers need not be slow.
> Starting with nothing compiled, on a project with over 17,000 source files
> (million-plus SLOC) ....
> On a Sun SPARCserver-1000, with one processor, and source files NFS
> mounted, and a fairly heavy load from other users (it's one of our main
> fileservers) .....
> I ran a script that does gnatmake with optimization for 98 executables.
> Took about ten hours. Takes _MUCH_ less (but I don't have exact
> numbers) on
> a four-processor Ultra-2 with source on local disks.
> Apex speed on the same source is slower but comparable.