home *** CD-ROM | disk | FTP | other *** search
- Path: sdd.hp.com!inn
- From: Jeff Grimmett <jgrimm@sdd.hp.com>
- Newsgroups: comp.sys.amiga.programmer
- Subject: Re: YASL (was: AsynchIO (was: fastest file read method ??))
- Date: 29 Jan 1996 19:50:05 GMT
- Organization: Hewlett-Packard Company
- Message-ID: <4ej8dd$hi1@news.sdd.hp.com>
- References: <w9YbsMD4ACazz9@jeff.dame.shnet.org> <w+RYXMD4FC8aRz1@_crisi.blackbox.shnet.org> <4dsalp$t68@news1.halcyon.com> <4du0ju$84b@news.sdd.hp.com> <4e2e76$9k3@toad.stack.urc.tue.nl> <4e3j8v$jd0@news.sdd.hp.com> <4e7si4$ore@tuegate.tue.nl> <4e8asc$rht@news.sdd.hp.com> <1349.6599T1259T2701@amiga.pp.se> <4egkuh$j1q@news.sdd.hp.com> <19960129.40EA00.9A87@am174.du.pipex.com>
- NNTP-Posting-Host: hpsdv330.sdd.hp.com
- Mime-Version: 1.0
- Content-Type: text/plain; charset=us-ascii
- Content-Transfer-Encoding: 7bit
- X-Mailer: Mozilla 1.2N (Windows; I; 16bit)
-
- m.hendry@dial.pipex.com (Mathew Hendry) wrote:
-
- >: I'd just like to see a little judgement used by developers, is all. If
- >: there's no real compelling reason to put it in a library, it shouldn't
- >: be. But, if you do, DOCUMENT IT and make the required files available to
- >: other developers so that it is NOT a waste.
-
- >So your judgement is that the AsyncIO routines should NOT be in a shared
- >library. Why is this?
-
- I thought I was sufficiently vague on that, but alas... no. My feelings
- on AsynchIO in a shared library is that if it were used sufficiently, it
- should be. At this point, I have downloaded and/or ftp'd a grand total
- of zero programs that require the shared library. I have one program
- that to my knowledge uses the linked library, but that's simply because I
- know the author.
-
- My MAIN reservation in this SPECIFIC case is that, as I understand it,
- AsynchIO was a CBM/CATS released archive. The archive I have does not
- contain a shared library. Either my version is old, or else the one with
- the shared library is bogus. I'm perfectly willing to admit the former,
- as the version I have required a fix to operate properly.
-
- >Several points which you should consider are:
-
- >1. It IS well documented.
-
- Yes, quite.
-
- >2. The library (AsyncIO.library) is tiny - under 3k.
-
- Never seen it. I'll take your word on it.
-
- >3. It would be a useful addition to many types of program (in fact, just about
- > any program which uses disk I/O).
-
- I've had it for 2 years. ONE program I've written made use of that, and
- I took it back out because the speed increase wasn't sufficient enough to
- be worth it.
-
- I won't deny point 3, just saying it hasn't demonstrated as much promise
- as it would appear to.
-
-
- >4. Due to point 3, it IS beginning to appear in many different programs. This
- > means that space savings (such that they are for a 3k library ;) and ease
- > of update are already becoming points in it favour.
-
- Who IS updating it? Are we talking about the same archive?
-
- >You might say that point 2 is a justification for using the link library
- >version instead, but that method requires a recompile for every update of the
- >library's functions. Wasted effort, if you ask me.
-
- Again, who's updating it? The archive I have was issued by CATS. They
- no longer exist and it doesn't appear that ATG's team is quite yet on the
- case.
-
-
- Slightly off the topic here, but maybe if it's a sufficient improvement,
- its routines should replace the current ones for file I/O in future revs
- of the OS.
-
-
-