). Any thoughts?
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
since I debugged the armlet initially as a 68K routine.)
But I don't know if anybody but other than me cares about the search speed.
Alex
--
Dr. Alexander R. Pruss
Department of Philosophy
Georgetown University
Washington, DC 20057-1133 U.S.A.
e-mail: [EMAIL PROTECTED]
online papers and home page
that the search be orderly.)
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
Bill,
If you look at the source of the class Mapper, in the file
PyPlucker/Writer.py, you will see the code which does the actual
record-id assignment.
Yes. I am looking at the fragment:
for (url, doc) in collection.items():
self._get_id_for_doc(doc)
_get_id_for_doc
Alternately, I suppose, I could add to the collection a field (or whatever
one calls it in Python) giving the time and date (or just a sequential
number) that the file got added in, and sort by that. I guess I might
just have to learn some Python. :-( Alex
--
Dr. Alexander R. Pruss || e-mail
. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
Does the python distiller also start thrashing under Linux with a lot of
files or is this a Windows-only problem? I am trying to distill a site
with 3000+ files, and is starts thrashing around file 1250.
There isn't a newer perl distiller, is there?
Alex
--
Dr. Alexander R. Pruss || e-mail
On Wed, 28 May 2003, David A. Desrosiers wrote:
What site are you hitting with the Python distiller?
A local copy (produced with wget) of newmanreader.org
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown
Alternately, one could just assign the key at the time that a file is
fetched. There would still, however, be the issue of the fragments of a
split-up file.
Alex
--
Dr. Alexander R. Pruss
Department of Philosophy
Georgetown University
Washington, DC 20057-1133 U.S.A.
e-mail: [EMAIL PROTECTED
on Linux. It can't
seem to be able to handle jpg files. I must be missing something.)
Alex
--
Dr. Alexander R. Pruss
Department of Philosophy
Georgetown University
Washington, DC 20057-1133 U.S.A.
e-mail: [EMAIL PROTECTED]
online papers and home page: www.georgetown.edu/faculty/ap85
be
wrong). I think the code should do a final check whether any of the links
in the files fetched at maximum depth can be resolved to things already
fetched. This would be nice.
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page
Problem is windows-specific. :-) I got the Linux version running.
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
On Thu, 29 May 2003, Michael Nordstrom wrote:
On Thu, May 29, 2003, Alexander R. Pruss wrote:
Thus, suppose I want to get only two URLs from my
home page: index.html and cv.html . Now, index.html links to cv.html and
to many other things. Ideally, I could have an include list of two URLs
depth calculated
from two different roots.
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
for free (i.e., without fetching more data from the web). The latter
feature would be rather nice, since it would allow one to control
precisely what order pages are put into the pdb.
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home
By the way, when new options get added, are people trying to keep the
prefs backwards compatible, and if so, how? (When I added the multiple
term search and transliteration search options, I used bitfields of an
existing pref item, the case-free search one.) Alex
--
Dr. Alexander R. Pruss || e
Still, considering the amount of code we are talking about (a few
lines) we could even make it possible for the user to select whether
the strike-through or bold font should be used for read documents.
Or, more precisely, whether strike-through should be used for read
documents, or bold for
) \
| ( unsigned char* )(x)[1] 8 ) \
| ( unsigned char* )(x)[0] )
#define SWAP16(x) ( ( (unsigned short)(x) 8 ) | ( (unsigned short)(x) 8 ) )
Should we just include that?
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL
?
2. How hard will it be to merge in the branch?
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
original use.
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
:
sed s/[bB][rR]/p/g
(I think the rest of the distiller doesn't mind just a p instead of
p.../p.)
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC
Thanks for the tip. The docs do say that SiteScooper does 1-, 2- and
3-level sites. That is rather limiting. Or does it do more? Even if it
does, I think there would be no harm, and some gain, in just including a
simple input pipe option in PyPlucker.
Alex
--
Dr. Alexander R. Pruss || e
compatibility, since some old values would no longer be valid.
Another option is this. For autoscroll speeds between 0 and 100ms, we
adjust by 10ms, and above that we go by 50ms.
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page
On Thu, 12 Jun 2003, Robert OConnor wrote:
Alexander R. Pruss wrote:
I've noticed that the autoscroll speed doesn't have enough
resolution. Currently, the minimum adjustment interval is 50ms. I'd be
tempted to change that to 10ms or 20ms, except then one would have the use
This sounds very good. I suggest we wait on this a little bit and apply the
requisite changes after the armlet code gets revamped after Mike reviews it.
I think we should keep the armlets/ directory because there may be more
armlets coming. Alex
___
wife that forces left justification when desired, unless
there is some way to fix this in the distiller.
--
Dr. Alexander R. Pruss
Department of Philosophy
Georgetown University
Washington, DC 20057-1133 U.S.A.
e-mail: [EMAIL PROTECTED]
online papers and home page: www.georgetown.edu/faculty/ap85
idea of just making TAB generate a
command stroke? (I tried it and it works fine.)
Alex
--
Dr. Alexander R. Pruss
Department of Philosophy
Georgetown University
Washington, DC 20057-1133 U.S.A.
e-mail: [EMAIL PROTECTED]
online papers and home page: www.georgetown.edu/faculty/ap85
What's the difference between the mainform and the fullscreenform?
--
Dr. Alexander R. Pruss
Department of Philosophy
Georgetown University
Washington, DC 20057-1133 U.S.A.
e-mail: [EMAIL PROTECTED]
online papers and home page: www.georgetown.edu/faculty/ap85
I wanted to know something more basic, but I think I figured it out from
your message. Fullscreenform is for viewing full screen images, right?
(I was wondering whether it is a version of mainform for viewing in what
Sony calls full-screen mode, 320x480 without a status line.)
Alex
Does every end of paragraph (except maybe the last in the record) have a
0x38 newline code after it in the plucker format?
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty
Summa Theologica. (It's not found :-) ) From memory, I think it
took about six to ten minutes. Right now, it does it in 1 min 49 sec with
the 8-bit optimized 68K search and 49 sec with the armlet search.
Alex
--
Dr. Alexander R. Pruss
Department of Philosophy
Georgetown University
Washington, DC
NativeFunction().
--
Dr. Alexander R. Pruss
Department of Philosophy
Georgetown University
Washington, DC 20057-1133 U.S.A.
e-mail: [EMAIL PROTECTED]
online papers and home page: www.georgetown.edu/faculty/ap85
I thought it might be easier to download a zip file of my fast search
Plucker. The zip is at www.georgetown.edu/faculty/ap85/dl/plucker.zip
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University
On Sun, 15 Jun 2003, Adam McDaniel wrote:
On Sun, Jun 15, 2003 at 09:06:51AM -0400, Alexander R. Pruss wrote:
Is there a way to specify the entry point to an armlet? With the way things
are currently set up, it looks like the armlet enters at the first function
in armlets/search.c
the first letter of the search string to 'a'). It was
Test() that got invoked, not NativeFunction().
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057
On Mon, 16 Jun 2003, Michael Nordstrom wrote:
On Sun, Jun 15, 2003, Alexander R. Pruss wrote:
Oh well. I was hoping that I could detect the end of paragraph /
end of line straightforwardly in the search code just while
scanning the text.
The record header includes everything you need
external font packages
with unencumbring licenses. :-)
--
Dr. Alexander R. Pruss
Department of Philosophy
Georgetown University
Washington, DC 20057-1133 U.S.A.
e-mail: [EMAIL PROTECTED]
online papers and home page: www.georgetown.edu/faculty/ap85
On Wed, 18 Jun 2003, John Marshall wrote:
On Tue, Jun 17, 2003 at 07:39:38AM -0400, Alexander R. Pruss wrote:
I've spun off fontconv.c into a separate project,
palmfontconv.sourceforge.net, so if you make changes to it, I would
appreciate it if you also sent a patch to me.
How much
of system fonts
should one allow for? People use fonthack, various localization software,
Fonts4NX, etc. How much extra room should one leave on forms in case
people change their system fonts in this way to something with wider
metrics?
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED
Changing the Prefs() function to a macro (and making the prefs structure
global) saves almost 2K. Is there any reason not to do it? Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu
You'll also find the binaries that cause the problem.
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
On Sun, 22 Jun 2003, Adam McDaniel wrote:
On Sat, Jun 21, 2003 at 09:50:38PM -0400, Alexander R. Pruss wrote:
What to do? I guess we need to promote a whole bunch of Int16s to
Int32s. This is going to degrade performance, but I see no cure. We need
to do this in TextContext
On Sun, 22 Jun 2003, Adam McDaniel wrote:
The back button is still clickable, regardless of appearing.. it just
so chooses to hide itself so the user can see whats under it :)
But there was nothing under it. :-)
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy
I committed a fix. Since the structure of the metadata has had to be
changed, Cleaner MUST be run to delete the temp files before running this
version. I don't know where to flag this in cvs.
Henceforth, please store y-offsets measured in pixels, except for things
that are on screen (or just
in half. (Initial experiments are not completely successful, because
I don't fully understand the two functions, but there is a great speedup.)
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University
about the speed of opening a new page.
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
On Sun, 22 Jun 2003, Adam McDaniel wrote:
There's a better way. Increment the MetaDocumentVersion in const.h
Committed. Thanks. Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu
with the subscript stuff. I hope it's not broken
now. There are imperfections in rendering in the large font, but they
match what was there in the previous version. One of these days, I might
fix it.
Please report any bugs!
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
. I just probably shouldn't be putting much more time into
this.
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
a couple of
times, with the Search in all pages option. The seek time is the time
from selecting an item in the search list, to that item showing up on the
screen. Is that an acceptable length of time?
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers
why the
effect would be non-uniform. Another thought is that the ARM is fast
enough here that something else is dominating.
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty
Is there any good reason why verticalOffset is always negative?
Alex
--
Dr. Alexander R. Pruss
Department of Philosophy
Georgetown University
Washington, DC 20057-1133 U.S.A.
e-mail: [EMAIL PROTECTED]
online papers and home page: www.georgetown.edu/faculty/ap85
.
But if there are complaints about the sign flip, I shan't do it.
Alex
On Wed, 25 Jun 2003, Adam McDaniel wrote:
On Wed, Jun 25, 2003 at 10:27:15AM -0400, Alexander R. Pruss wrote:
Is there any good reason why verticalOffset is always negative?
Its becuase its the relation of the top of the page to the top
Here's how to test it. Search for some word that occurs a couple of
times, with the Search in all pages option. The seek time
is the time from selecting an item in the search list, to that item
showing up on the screen. Is that an acceptable length of time?
I'm using hires
From: Adam McDaniel [EMAIL PROTECTED]
On Wed, Jun 25, 2003 at 03:12:08PM -0400, Alexander R. Pruss wrote:
But if there are complaints about the sign flip, I shan't do it.
Well, my only concern is that it keeps the context of the *Offset
values with each other. However it might be redundant
Here's an interesting bug. Under POSE/NR70, minimizing the silkscreen
also triggers the GESTURE_TAP gesture sometimes. (I think only the first
time it's done during a session.)
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page
Bookmarks, history and current position are all now supposed to be
maintained across font changes. I committed this to cvs. Alas, one will
lose old bookmarks if one upgrades. Oh well: it does let me resolve two
bugs.
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy
with documents in RAM, and haven't yet tried on MS.)
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
, on Sony NX/NR, we'll want to toggle the
status bar, too), which I've added.
Let me know about any bugs.
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057
Skins??
Yeah. Drop a prc with creator code Plkr and resource type Skin, and
resources there (e.g., alternate renditions of icons) become available.
However, you need to make sure that these resources aren't already in
Plucker, so you might need a custom build (or to delete the resources).
Not
POSE/NR70 reports a write to ROM (in fact, several of them) in every
dialog window. (E.g., search in page for a non-existent word ad get the
write to ROM at the String not found dialog.) Should we be worried?
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department
.
Alex
p.s. I've noticed that a bunch of things in util.c don't conform to the
coding guidelines. But who am I to speak?
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
, which is messy.
if( condition1
condition2 ) {
ConditionSuccessful();
}
This is more readable but messy.
if( condition1
condition2
) {
ConditionSuccessful();
}
This seems OK, but the ) { line looks funny.
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL
underground? I am worried when bugs disappear
between builds with no one having fixed them.
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
I've been looking at the library jogdial code, and noticed that a function
uses UInt8 to store the row. Now, I haven't figured out quite how the
code works, but is there a 255 document limit, or is this just an offset
within the page?
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED
pretty-printing so that the programmer can fix it himself).
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
I need a document, ideally a .pdb, that I can test Plucker's unicode
handling on. Does anyone have such a document? Alex
--
Dr. Alexander R. Pruss
Department of Philosophy
Georgetown University
Washington, DC 20057-1133 U.S.A.
e-mail: [EMAIL PROTECTED]
online papers and home page
. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
in
external font packages (not that we're distributing any, but I do have a
cyrillic package if anybody wants one).
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC
iSilo is the viewer. It has a free mode in which hyperlinks and various
other things don't work./ Otherwise, one pays. iSiloX is the
distiller. It's free.
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University
too much Heck, I've been using the
Unofficial Hires version since last November
Well, if I am not mistaken, it's all in HEAD in CVS now, so as soon as the
daily snapshot make is fixed, it will be available daily on plkr.org, no?
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED
for the text to display
correctly.
So there are character set records in the metadata section? That's very
nice--I should be able to make use of this somewhere in the font support
code.
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home
(). But TxtGlueGetNextChar(), I expect, slows things
down. If the answer to #2 is negative, then this is easy--I just scan
ahead character by character to a null.
I might just have to make some alternate code for 8-bit encodings.
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED
correct behavior, because it will allow
line-breaking at any two-byte length character.
2. Thus, how do we check that the character is indeed CRLF? What is the
two-byte encoding of CRLF?
3. And why should there be a CRLF in the record given that we have
function 0x38 instead?
Alex
--
Dr. Alexander R
One of my recent optimizations involved using FntWordWrap() for some of
the rendering. But now I have a worry. Maybe FntWordWrap() can't handle
soft hyphens? Does anyone know?
Also, does anyone know what FntCharsWidth() returns if there is a soft
hyphen in the string?
Alex
--
Dr. Alexander R
. Or am I misunderstanding something in the code?
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
out by the distiller.
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
character to the
left of the last soft hyphen or space, and then run things through the
standard GetLineMetrics() handler.) Perhaps text that does a lot of soft
hyphen breaking will be processed a touch slowly, but it should render OK
(though I haven't tried it).
Alex
--
Dr. Alexander R. Pruss || e
Using the general prefs settings to change to depth 4 doesn't work? Alex
--
Dr. Alexander R. Pruss
Department of Philosophy
Georgetown University
Washington, DC 20057-1133 U.S.A.
e-mail: [EMAIL PROTECTED]
online papers and home page: www.georgetown.edu/faculty/ap85
2. Can multi-byte encoded texts have nulls that do not signal the end of
the string, i.e., nulls in the middle of multi-byte characters? (I
assume
there can't be nulls at the start of them--if there can, our code is in
trouble.)
Uhmm.. although this isn't my area of expertise, but I
On Tue, 8 Jul 2003, Michael Nordstrom wrote:
On Tue, Jul 08, 2003, Alexander R. Pruss wrote:
1. This looks to me like it's checking for all two-byte length characters,
not just CRLF. This surely isn't correct behavior, because it will allow
line-breaking at any two-byte length character
On Tue, 8 Jul 2003, Michael Nordstrom wrote:
On Tue, Jul 08, 2003, Alexander R. Pruss wrote:
Nevermind. I have a horrible memory for numbers--I misremembered the
0x00AD as a 0x00A5.
I think we should mind anyway. Bug report #356 is a related problem
for 0xAD and the Thai language.
We
On Tue, 8 Jul 2003, Michael Nordstrom wrote:
On Tue, Jul 08, 2003, Alexander R. Pruss wrote:
I am also a bit puzzled as to why the special handling for
two-byte characters? (Why not three- or four-byte ones?)
Because that's what Matto implemented... If you want to see something
else
Is there anywhere one can ask to get an official answer to such a
question, as my question whether middle or final bytes can be zero?
--
Dr. Alexander R. Pruss
Department of Philosophy
Georgetown University
Washington, DC 20057-1133 U.S.A.
e-mail: [EMAIL PROTECTED]
online papers and home page
What is multibyte support about then? I found a reference to this in the
archives, but it seems this is available in the I18N branch only.
Unfortunately I still can't reach www.plkr.org or cvs.plkr.org to check
this out.
Multibyte support is for devices, e.g., Japanese ones (I think these are
On Tue, 8 Jul 2003, Bill Janssen wrote:
So there is no support for 0xA5 in the code (and nothing to remove);
This is true for 0xA5. But it's not true for 0xAD. Support for the
soft-hyphen is hardcoded in.
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department
at the beginning of such a character will terminate
things. (At least for the parts of the code that I've looked at in more
detail, i.e., paragraph.c and search.c.)
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University
);
return true;
} ErrEndCatch
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
dumb in the patch.
Note:
When I say work, I mean work while keeping the order right. As
far as I can tell, my code will not fail in any really bad way even if the
limits are exceeded--the documents will just be out of order.
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED
agree it would be nice.
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
binary search over the index range max(1,K-(M-N)) to min(K,1+N), which
will at least sometimes be a smaller range than 1 to K. (If the ids are
assigned sequentially, then in fact we will hit right on the index!)
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department
On Fri, 11 Jul 2003, Michael Nordstrom wrote:
On Fri, Jul 11, 2003, Alexander R. Pruss wrote:
Currently we do a nice binary search for ids when looking for a record.
But we can do a little better than that.
Sure, but do we really *have to*, i.e. will you even notice the
difference
A better solution is to make the spacing between the document file ids
equal to (65536-10)/numitems. Any more suggestions? Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty
under Windows, and a second time under Linux, so it's not
an OS issue.)
Does PyPlucker maybe keep all the texts in RAM rather than saving them to
a cache? Can this be fixed?
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page
to keep the order straight. My mods seem
to work, but need someone who knows Python to have a closer look at them.
(I've got an updated patch if anybody wants to look at it.) If nobody
volunteers, I guess I'll just use the patched version myself and see how
it goes.
Alex
--
Dr. Alexander R
.) This seems to be the thing to
do if one wants to fix the RAM problem.
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
fetched rather than alphabetically.
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
to dot is active). Oh well.
Alex
--
Dr. Alexander R. Pruss || e-mail: [EMAIL PROTECTED]
Philosophy Department || online papers and home page:
Georgetown University || www.georgetown.edu/faculty/ap85
Washington, DC 20057||
U.S.A
It looks like it's really VERY easy to get the distiller to do what I
want. I just sort the collection before the initial round of id
generation, and then I loop over subdocs and not just main docs. And that
does it. Minimal code modification, and does the job.
Alex
--
Dr. Alexander R
1 - 100 of 561 matches
Mail list logo