Re: A85: Rigel, where'd it go??


[Prev][Next][Index][Thread]

Re: A85: Rigel, where'd it go??



On Tue, 14 Oct 1997 19:18:50 -0400, you wrote:

>Jimmy Mårdell wrote:
>> As Sam & Andi has mentioned, it's one byte most of the time. The format
>> is very simple: if the next byte in the relocation table is >0, it's
>> the relative address from the previous relocation address. If it's 0,
>> the next word in the table is the relative address from the previous
>> relocation address.
>> 
>> Personally, I've never liked fixed address relocation because it
>> seems very unstable, especially if you're going to support libraries,
>> TSR, deleting/resizing vars etc. Deleting a variable between
>> the fixed program and libraries in use (+ TSR programs running) must
>> require quite a lot of code since you have to rerelocate the library
>> and the TSR program.
>
>Maybe that's been the case in your attempts with Usgard.  However, Rigel
>1.0, which supports Fixed Address Relocation, is very stable.  One can
>have TSRs, libraries, and use the Var functions (I've tested).  And as
>for it taking a lot of code.  Well, somehow I managed to get Rigel in a
>size 800 bytes less than Usgard.  So, do you have a different opinion on
>fixed address relocation?
>

I think that you are over estimating the size of Usgard.  With a shell
similar to the one for rigel, it is about 1800 bytes.  Rigel, without
all of the features you talked about (i.e. v0.9) is 1200 bytes.

-mike pearce


Follow-Ups: References: