i have a TSortedCollection that i'm allowing duplicate keys to be inserted in... the problem that i'm having is that the duplicate keys are in reverse order from how they were inserted...

i am overriding MyCollection^.Insert because i have additional tasks that need to be done at the time of insertion... without duplicate keys, everything works perfectly so that lesser keys do have a lower index than higher keys... with duplicate keys, though, the lesser keys are right but the duplicate lesser keys are in reverse order within their "grouping"...

MyCollection^.Insert uses AtInsert to insert the new record... i switched this to Insert after i read the documentation statement that AtInsert should not be used... when i did this, my program crashed with an unknown runtime error 202 (stack overflow???)... at a guess, i would say that i ended up calling MyCollection^.Insert recursively... perhaps instead of just AtInsert i should use something_else.Insert inside my Insert??

anyway, how can i fix this problem where duplicates are inserted in the proper order while still retaining the existing proper key ordering?

waldo kitty wrote

i have a TSortedCollection that i'm allowing duplicate keys to be inserted in... the problem that i'm having is that the duplicate keys are in reverse order from how they were inserted...

Sounds like your TSortedCollection.Compare() needs additional logic. It isn't doing enough comparisons internally to order things properly.

i am overriding MyCollection^.Insert because i have additional tasks that need to be done at the time of insertion... without duplicate keys,

Wouldn't it be better to perform your additional tasks before the insertion, then call a normal .Insert? Why overload something if there isn't really a need to?

possibly... i dunno... currently it compares the (pstring) keys only... i thought about trying to also include a second field (a double) which is used for the "time stamp" but i don't know how to do that without having to use the key and this "time stamp" field every time i do anything with the records...

i kinda thought about that earlier when i was digging thru the code... IIRC, insert was overridden because there are cases where the existing record needs to be replaced (because the new record is newer) using AtPut and the non-added records must be logged and their reason for not being added (older or same)...

2014-02-04 waldo kitty :

Just a quick idea which could be completely wrong, but wouldn't it solve your issue if Compare always returned -1 instead of 0?

Frederic Da Vitoria wrote

2014-02-04 waldo kitty < >:

[...]

no, because i lose (at least) the logging of why the record was tossed out... see below... it wouldn't help with the duplicate keys being inserted in reverse order...

Procedure TTLEColl.Insert(Item: Pointer);
var
   I   : Sw_Integer = 0;
   old : Pointer;
   foo : integer = 0;
begin
   I := 0;
   If NOT Search(KeyOf(Item), I) OR Duplicates Then  // Item valid
     begin
       logTLEEntryType('M');                         // MERGE this TLE
       AtInsert(I, Item);                            // Insert the item
//      Insert(Item);                                 // Insert the item
       inc(cur_TLEs_added);                          // increment the add counter
     end
   else                                              // otherwise
     begin
       foo := CompareReal(PTLERec(At(i))^.epoch,PTLERec(Item)^.epoch); //
compare the epochs
       case foo of
         -1 : begin                                  // replace existing record
                logTLEEntryType('U');                // UPDATE this entry
                old := At(i);                        // save existing pointer first!
                AtPut(i, Item);                      // now put in the new record
                dispose(PTLERec(old),done);          // dispose old one
                inc(cur_TLEs_updtd);                 // increment the updated
counter
              end;
          0 : begin                                  // we're tossing this one out
                logTLEEntryType('S');                // this entry is the SAME
                dispose(PTLERec(Item),done);         // dispose the item
                inc(cur_TLEs_same);                  // increment the same counter
              end;
          1 : begin                                  // we're tossing this one out
                logTLEEntryType('N');                // the existing entry is NEWER
                dispose(PTLERec(Item),done);         // dispose the item
                inc(cur_TLEs_old);                   // increment the old counter
              end;
         end; // case
     end;
end;

2014-02-05 waldo kitty :

Hm, yes, you don't always want to keep duplicates.

it wouldn't help with the duplicate keys being inserted in reverse order...

Once again I did not test this, but it seems to me that if Compare returned -1 instead of 0, any duplicate would be inserted after because it would never be considered as equal to any other. But since you still want your collection to be able to choose between skipping duplicates or keeping them, the Compare modification would have to be slightly more subtle, something like:

if result = 0 and Duplicates
    then result := -1

at the end of the Compare function.

On 2/5/2014 3:57 AM, Frederic Da Vitoria wrote:
[...]

i tried this and it kinda works... it keeps the entries in their original order but...

1. the final logging of each item's "record number" (position in the collection) is -1
2. it doesn't help to sort them in order by a different field

i'm unsure what to do or how to handle this so that there is a secondary (sub) sorting order so that the main key is the master sort and then a secondary ""key"" is used when duplicates are allowed... ideally, the secondary key would retain the original order in the case that the ""secondary key"" is exactly the same as a previous ""secondary key""... but this is also problematic...

to try to clarify: sometimes there are records released with exactly the same time stamp (epoch in my code i posted) but slightly different data within the record... there is another field that might be used to differentiate those BUT the records come from numerous locations... they may or may not use this ""tertiary key" and if they do, their numbering in this ""tertiary key"" may not be the same as any other system's count for this ""tertiary key""... this is a problem i don't know how to solve as there is no coordination between locations and no ""master"" coordinator for this ""tertiary key""... it becomes even more apparent because my flow doesn't take any certain record containers before any others... they are read and processed as they appear (OS ordering actually) which may cause ""newer"" records with an identical time stamp to be processed after others... in my current design, i'm using "first come, first served" meaning that the first record processed is the one that is retained... with dupes, this doesn't matter so much but it does all still come into play...

waldo kitty wrote

i'm unsure what to do or how to handle this so that there is a secondary (sub) sorting order so that the main key is the master sort and then a secondary ""key"" is used when duplicates are allowed...

You put both key comparisons in the same .Compare function.

In one of my Turbo Pascal projects, I used this:

PSystems=^TSystems;
TSystems=object(TSortedCollection)
   function Compare(Key1, Key2: Pointer): Integer; virtual;
   function GetSize:longint;
end;

...

Function TSystems.Compare;
begin
   if PSystem(Key1)^.score < PSystem(Key2)^.score
     then Compare := -1
     else if PSystem(Key1)^.score > PSystem(Key2)^.score
       then Compare := 1
       {Score is the same.  Second search key (larger numbers are worse:}
       else if usectotal(Key1) > usectotal(Key2)
         then Compare := -1
         else if usectotal(Key1) < usectotal(Key2)
           then Compare := 1
           {usec totals same.  Third search key:}
           else if PSystem(Key1)^.BIOSCRC16 < PSystem(Key2)^.BIOSCRC16
             then Compare := -1
             else if PSystem(Key1)^.BIOSCRC16 > PSystem(Key2)^.BIOSCRC16
               then Compare := 1
               else Compare := 0;
end;

2014-02-06 waldo kitty :

Then my trick does not work for you because it hides the fact that the records are identical. You need to give to someone the responsibility of giving the secondary key. If I understand what you wrote correctly, there is already something which could be used as a tertiary key, but it doesn't really work because the way it is filled is not consistent across the different sources. If I were you, I'd keep this tertiary key data (I guess it is meaningful, so you can't remove it), and I'd create my own secondary key inside the TSortedCollection descendant. I'd use 2 compare functions,
- one which works as your current one but wouldn't be declared as a compare function (you could call it CheckPrimaryExists) and which would return 0 if the primary key already exists
- and one which uses both the primary and the secondary key as Jim suggested.

The algo (when duplicates are allowed) would be something like: if primary key exists
then set secondary key to a number
insert the data

Note that depending on the total number of rows, you could use a general counter for the secondary key, no need to fetch the value of the last secondary key for the same primary key.

You'd get something like (primary key / secondary key) A / 0
B / 0
B / 1 (duplicate detected, first secondary key generated) C / 0
C / 2 (duplicate detected, second secondary key generated, note that there is no C1)
...

... and now that I think of it, you don't need 2 compare functions, the second one should work for both usages.

compiling for targets win32 and win64, I´ld like to change the window title, the window opened by cmd.exe

program x;
uses ?;
begin
  yyyyy( 'MY TITLE');
end.

I didn´t find anything in the windows unit ....

Thanks for help!
Philippe

2014-02-06 Philippe :

I don't understand the relation between cmd.exe and your program source. Do you mean that you want to create a program which would open a cmd.exe window and give it a title? Or do you want to run cmd manually and from inside it run a program which would change the window's title? Or something else?

searching for raw printing to usb printer ... I found that some stuff has been posted in the community forum ... which looks to be out of order ... is there a way to get the informations posted ...

or may
be someone could tell me how to raw print to usb (as for LPT1), for exemple to Epson LX 80.

I am using fpc 2.6.2 targetting win32 and
win64.

(with Lazarus it seems it can be done using printer unit ... can I use it "directly"? )

Thanks for all

Philippe

Google suggests that Windows.SetConsoleTitle should do it (https://www.google.com/search?q=MSDN+change+console+window+title).

Tomas

just simple like that ... works perfectly as I need ...

I had
searched (but searching "freepascal") and found something ... more complicated ...

thank´s a lot.

Philippe

Tomas Hajny wrote

On Thu, February 6, 2014 15:16, Philippe wrote: > compiling for targets win32 and win64, I´ld like to change the

window title, the window opened by cmd.exe program x; uses ?; begin yyyyy( 'MY TITLE'); end. I didn´t find anything in the windows unit ....

Google suggests that Windows.SetConsoleTitle should do it

(https://www.google.com/search?q=MSDN+change+console+window+title [1]).

fpc-pascal maillist -

http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal [2]

Links:
------
[1]
https://www.google.com/search?q=MSDN+change+console+window+title [2]
http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal

interesting...

ahh... i see what you are doing... you are passing the pointer to the whole record instead of just the needed key... hummm... i'll have to think about that and how to do it in my app... right now i'm only passing the necessary key because the record is 'large'... but then again... some thought and meditation is necessary ;)

Frederic Da Vitoria wrote

Then my trick does not work for you because it hides the fact that the records are identical.

actually, the problem comes when referencing the IndexOf() the record... IndexOf(Item) is passed (unmodified) thru a couple of routines to finally end up here as 'idx'... IndexOf() being the TSortedCollection IndexOf() routine...

Procedure LogTLEPerApo(tleRecord: PTLERec; idx: sw_integer; var oFile: text);

begin
WriteLogLine(AppLogFile,0,
PadLeft(IntToStr(idx),5)
+ ' ' + PadRight(tleRecord^.catnbr^,5)
+ ' ' + PadRight(tleRecord^.cospar^,8)
+ ' ' + PadRight(tleRecord^.satname^,27)
+ ' ' +
PadRight(AddChar('0',FloatToStrF(tleRecord^.epoch,ffFixed,15,8),14),15)
+ PadRight(displayEpochString(tleRecord^.epoch),19)
+ ' ' + PadLeft(FloatToStrF(tleRecord^.Perigee,ffFixed,15,8),16)
+ ' x ' + PadLeft(FloatToStrF(tleRecord^.Apogee,ffFixed,15,8),16)
+ ' x ' + PadLeft(FloatToStrF(tleRecord^.Inclination,ffFixed,15,8),16)
+ ' ' + PadLeft(FloatToStrF(tleRecord^.SMajAxisKM,ffFixed,15,8),16)
+ ' ' + PadLeft(FloatToStrF(tleRecord^.MMPeriod,ffFixed,15,8),16)
+ ' ' + PadLeft(FloatToStrF(tleRecord^.SMAPeriodC,ffFixed,15,8),16)
+ ' ' + PadLeft(FloatToStrF(tleRecord^.SMAPeriodP,ffFixed,15,8),16)

   );
   WriteOUTFile(tleRecord,oFile);
   inc(tot_TLEs_written);

end;

You need to give to someone the responsibility of giving the secondary key.

that's not going to happen... anyone can generate these records and make them available to others for use... i'm aware of at least 5 different groups and several owners of the objects who are generating these...

If I understand what you wrote correctly, there is already something which could be used as a tertiary key, but it doesn't really work because the way it is filled is not consistent across the different sources.

right... for the same reason as above...

If I were you, I'd keep this tertiary key data (I guess it is meaningful, so you can't remove it),

right... it is part of the format... some records don't even have this tertiary key filled in...

and I'd create my own secondary key inside the TSortedCollection descendant.

hummm... one group does something similar but i currently don't have any method of doing this... in all actuality i should probably be doing this in a sql database but the overhead of processing speed and storage space is more than i want to deal with... especially since a full processing run using all options takes less than a minute for ~40000 records and a previous tool (limited to 64k) takes almost 30 minutes...

definitely something for me to contemplate on...

this is one of the things i like about talking about problems... while others may not be fully aware of the data, they can still come up with ideas on how to handle situations... one of the things i used to do was to talk to a mechanic friend... i'd convert my tasks and problems to something vehicle related so that he could see it from his mechanic's point of view and then he'd ask about doing it like "this" in vehical mechanic terms which i'd convert back to programming and suddenly have a workable solution... he helped me solve more problems in this way than i can count and he barely knew where the "big red switch" was on the computers ;) ;) ;)

Philippe wrote

compiling for targets win32 and win64, I´ld like to change the window title, the window opened by cmd.exe

please do not hijack other people's topics to ask an unrelated question... use your 'write' or 'new message' button instead of 'reply' to start a your new topic... thanks!

Philippe wrote

searching for raw printing to usb printer ... I found that some stuff has been posted in the community forum ... which looks to be out of order ... is there a way to get the informations posted ...

and you did it (hijacked the same thread with another topic) again :/

Not much thought required -- a pointer to the record is 4 bytes, no matter how large the record is. In fact, it's probably faster than what you're doing, and more flexible.

sorry ... I´ll do it next time ...

waldo kitty wrote

On 2/6/2014 9:16 AM, Philippe wrote: > compiling

for targets win32 and win64, I´ld like to change the window title, the window opened by cmd.exe

please do not hijack other people's topics

to ask an unrelated question... use

your 'write' or 'new message'

button instead of 'reply' to start a your new

topic... thanks!

so ... sorry again (I didn´t know it could make a difference) ...

waldo kitty wrote

On 2/6/2014 9:24 AM,

Philippe wrote

> searching for raw printing to usb printer ... I

found that some stuff has been posted in the community forum ... which looks to be out of order ... is there a way to get the informations posted ...

and you did it (hijacked the same thread with another

topic) again :/

It depends on the reader's mail client. Each mail has a unique ID. When you chose to answer to a mail, your answer contains the ID of the original mail too. If you change the subject, this ID chain remains. This allows some mail clients (for example Thunderbird) to show the mails in a tree structure, like on a forum.

2014-02-06 Philippe :

thanks ... I´ll take care of it ...

Frederic Da Vitoria wrote

It depends on the reader's mail client. Each

mail has a unique ID. When you chose to answer to a mail, your answer contains the ID of the original mail too. If you change the subject, this ID chain remains. This allows some mail clients (for example Thunderbird) to show the mails in a tree structure, like on a forum.

again (I didn´t know it could make a difference) ...

> > Em

waldo kitty wrote

> >> On 2/6/2014 9:24 AM,

Philippe wrote

>> >>> searching for raw printing to usb printer ...

I found that some stuff has been posted in the community forum ... which looks to be out of order ... is there a way to get the informations posted ...

>> >> and you did it (hijacked the same thread with

another topic) again :/

> >

yep! i realized this when i was going back thru the code cleaning it up and letting the other side of my brain wander about in its own world ;)

In fact, it's probably faster than what you're doing, and more flexible.

i was sending the pointer to field's contents ;)

[time passes]

well, i thought i was... my KeyOf routine does a cast using

PTLERec(Item)^.catnbr

but my (TLE) compare is apparently only sending the pointer to the catnbr string because the cast there is simply

PString(Key1)^
and
PString(Key2)^

the only place i understand that my (TTLEColl) compare is in play is in

If NOT Search(KeyOf(Item), I) OR Duplicates Then // Item valid

from the TTLEColl.Insert routine override that i posted previously... that Insert routine is taking a pointer to an entire record...

so now i'm a bit confused but it has been a long day... again... :/

Philippe wrote

so ... sorry again (I didn´t know it could make a difference) ...

thanks and no problem... frederic's explanation is accurate... one should also think about news groups and their readers... the Message-ID and References lines of each post are used to build the topic trees and the links between messages to create the threading... this is much better than the old "subject linking" style that was (still is in some cases) used on the old original BBSes...

i do apologize if i sounded harsh or ugly... this happens more than you might think ;)

2014-02-07 waldo kitty :

This makes sense... at least in your previous version: I guess catnbr is a string. Compare uses whatever KeyOf returns, and since KeyOf returns the address of a string, Compare handles the keys as such. To compare more than 1 field, the simplest way would be for KeyOf to return the address of the TLERec, which is the behavior of TSortedCollection.KeyOf http://lazarus-ccr.sourceforge.net/docs/rtl/objects/tsortedcollection.keyof.html. So IMO you should simply remove your implementation of KeyOf and edit Compare so that it receives PTLERec and uses the relevant fields instead of a simple string.

no problem ... may be could be informed at the messages bottom ... "new topic: create a new message ..."

waldo kitty wrote

On 2/6/2014 2:45 PM, Philippe wrote: > so ... sorry

again (I didn´t know it could make a difference) ...

thanks and no

problem... frederic's explanation is accurate... one should also

think about news groups and their readers... the Message-ID and References lines

of each post are used to build the topic trees and

the links between messages to

create the threading... this is much

better than the old "subject linking" style

that was (still is in

some cases) used on the old original BBSes...

i do apologize if i

sounded harsh or ugly... this happens more than you might

Philippe wrote

>> >>> searching for raw printing to usb printer

... I found that some stuff has been posted in the community forum ... which looks to be out of order ... is there a way to get the informations posted ...

>> and you did it (hijacked the same thread

with another topic) again :/

2014-02-07 waldo kitty :

Sadly at least some popular webmail clients (GMail) don't use Message-ID either, so that the messages are in the order they were received instead of the order of the answers and some answers are shown in a separate thread because GMail thinks the subject has changed ("Re." is sometimes translated in strange ways).

Well, this should always be the behavior, not only for this mailing list. We could remind this, but the reminder should explain that this is not only for this ML but for all MLs and even standard mail usage. For example, I usually use the threaded view in Thunderbird for all my mails. This means that if we start a personal conversation you and me (not through the ML) and if at some point you change the subject instead of creating a new mail, in my Thunderbird client, your "changed subject" answer will still be in the first thread, which probably was not what you intended.

2014-02-07 Philippe :

Frederic Da Vitoria wrote

Well, this should always be the behavior, not only for this mailing list.

In the era of facebook and twitter communication habits are bound to decline.

-Michael

Frederic Da Vitoria wrote

Sadly at least some popular webmail clients (GMail) don't use Message-ID either,

Posts without a Message-IDs should be discarded (and sent back) by the mailing list server !
(OK, I sometimes send an additional personal message to the original poster by hitting the wrong button in Thunderbird :-[ )

-Michael

2014-02-07 Michael Schnell :

Of course, but we can try to explain that twitter and facebook usage are for twitter and facebook and don't necessarily apply elsewhere. Even users made brain dead by twitter and facebook should be able to understand this :-) You don't drive the same way in a city and in a motorway. You don't post the same way using mail or twitter or facebook. Even if only a dozen users finally understand what we are saying and why we do it this way, it is worth the effort IMO.

[snip]

Of course, but we can try to explain that twitter and facebook usage are

[snip]

Please always post such off-topic comments to the fpc-other list.

Thanks.

Jonas
FPC mailing lists admin

Frederic Da Vitoria wrote

2014-02-07 waldo kitty < >:

[...]

yes...

i've read that numerous times but...

So IMO you should simply remove your implementation of KeyOf and edit Compare so that it receives PTLERec and uses the relevant fields instead of a simple string.

seriously? [/rhetorical] son of squidly! i don't know why but i've always overridden keyof as well as compare and at least one other that i can't think of at the moment... now that i've completed converting my "decision tree" to using two bits per option instead of the three-state format i was using (the exe was reduced by a few 10's of k's) i will see about working on this aspect...

while my app is in production and operating 24x7 on my systems, i still consider it to be no more than beta stage... personally, alpha but then again, i have pretty high testing standards for my code... while it might work, there's no way that i would allow it out in a commercial operation without a lot more testing and evaluation...

anecdote: some years back i wrote a messaging app to the specs for the use it was to be placed in... those specs stated that a message identifier would not be used within a three year period... testing of that code took just over three years to ensure that it was to spec... running mathematical algorithms showed that the code was up to spec but i preferred to use actual testing to make sure... i'm my own worst critic ;)