Parallel predictive compression and access of a sequential list of executable instructions

A method for decompressing segments of data as needed and without requiring sequential selection or decompression of the data even though the compression was performed using predictive models, the method comprising the following:

• an act of accessing a compressed version of a first segment of the data, the first segment being represented by a sequence of symbols;
• an act of decompressing the first segment of the data;
• an act of accessing a first data decompression predictive model which operates by predicting characters or symbols based on sequential processing of immediately preceding characters or symbols, wherein the first data decompression predictive model was generated using the sequence of symbols that represents the first segment;
• an act of determining that a second segment of the data is desirable, wherein the act of determining that a second segment of the data is desirable comprises the following:
• an act of executing an executable list within the first segment;
• while executing within the first segment, detecting that an instruction is needed from the second segment, wherein the act of detecting that an instruction is needed from the second segment comprises the following:
• an act of detecting an instruction in the first segment to go to an address in the second segment;
• an act of detecting the final instruction in the first segment;
• an act of accessing a compressed version of the second segment of the data;
• an act of decompressing the second segment of the data after decompressing the first segment, and using the first data decompression predictive model to predict characters or symbols in the second segment of data, wherein the first data decompression predictive model acts as a predictive starting point such that when decompression of the second segment of data begins, the first data decompression predictive model is based only on the first segment of data, such that when decompression of the second segment of data begins, predictions of characters or symbols of the second segment of data, which is not sequential with the first segment, are based entirely on the first predictive model of the first segment of data, and exclude consideration of all intervening segments of data between the first segment of data and the second segment of data;
• an act of determining that a third segment of the data is desirable;
• an act of accessing a compressed version of the third segment of the data;
• and
• an act of decompressing the third segment of the data after decompressing the first segment, and using the first data decompression predictive model to predict characters or symbols in the third segment of data, wherein the first data decompression predictive model acts as a predictive starting point, wherein the first data decompression predictive model is a common starting point for decompressing both the second segment and the third segment of the data, such that when decompression of the third segment of data begins, predictions of characters or symbols of the third segment of data are based entirely on the first predictive model of the first segment of data, and exclude consideration of the second segment of data as well as any intervening segments of data between the first segment of data and the third segment of data.

A computer program product for decompressing segments of data as needed and without requiring sequential selection or decompression of the data even though the compression was performed using predictive models, the computer program product comprising one or more computer-readable physical storage media have stored thereon the following:

• computer-executable instructions for causing a compressed version of a first segment of the data to be accessed, the first segments being represented by a sequence of symbols;
• computer-executable instructions for decompressing the first segment of the data;
• computer-executable instructions for causing a first data decompression predictive model that was generated using the sequence of symbols that represents the first segment to be accessed, the first data decompression predictive model operating by predicting characters or symbols based on sequential processing of immediately preceding characters or symbols;
• computer-executable instructions for determining that a second segment of the data is desirable, wherein the act of determining that a second segment of the data is desirable comprises the following:
• an act of executing an executable list within the first segment;
• while executing within the first segment, detecting that an instruction is needed from the second segment, wherein the act of detecting that an instruction is needed from the second segment comprises the following:
• an act of detecting an instruction in the first segment to go to an address in the second segment;
• an act of detecting the final instruction in the first segment;
• computer-executable instructions for causing a compressed version of the second segment of the data to be accessed;
• computer-executable instructions for decompressing the second segment of the data using the first data decompression predictive model to predict characters or symbols in the second segment of data, wherein the first data decompression predictive model acts as a predictive starting point such that when decompression of the second segment of data begins, the first data decompression predictive model is based only on the first segment of data, such that when decompression of the second segment of data begins, predictions of characters or symbols of the second segment of data, which is not sequential with the first segment of data, are based entirely on the first predictive model of the first segment of data, and exclude consideration of all intervening segments of data between the first segment of data and the second segment of data, and wherein decompressing the second segment occurs after decompressing the first segment;
• computer-executable instructions for determining that a third segment of the data is desirable;
• computer-executable instructions for causing a compressed version of the third segment of the data to be accessed;
• and
• computer-executable instructions for decompressing the third segment of the data using the first data decompression predictive model to predict characters or symbols in the third segment of data, wherein the first data decompression predictive model acts as a predictive starting point such that when decompression of the third segment of data begins, the first data decompression predictive model is based only on the first segment of data, such that when decompression of the third segment of data begins, predictions of characters or symbols of the third segment of data are based entirely on the first predictive model of the first segment of data, and exclude consideration of the second segment of data as well as any intervening segments of data between the first segment of data and the third segment of data, and wherein decompressing the third segment occurs after decompressing the first segment and after decompressing the second segment.

A method for compressing data, the method comprising the following:

• an act of compressing a first segment of the data using a sequence of symbols that represents the first segment;
• an act of generating a first data compression predictive model which operates by predicting characters or symbols based on sequential processing of immediately preceding characters or symbols, and using the sequence of symbols that represents the first segment;
• an act of compressing a second segment of the data using a sequence of symbols that represents a second segment of the data, and using the first data compression predictive model as a predictive starting point such that when compression of the second segment of data begins, the first data compression predictive model is based only on the first segment of data, such that when compression of the second segment of data begins, which is not sequential with the first segment of data, characters or symbols used for predicting, the second segment of data are based entirely on the first predictive model of the first segment of data, and exclude consideration of all intervening segments of data between the first segment of data and the second segment of data, wherein the act of compressing the second segment of the data is performed after the act of compressing the first segment;
• wherein the act of compressing a second segment of the data using a sequence of symbols that represents a second segment of the data, and using the first data compression predictive model as a predictive starting point comprises the following:
• an act of progressing the first data compression predictive model while compressing the second segment to thereby generate a second predictive model after compressing the second segment;
• an act of generating a log that describes the transformations imposed on the first data compression predictive model as it changes to the second predictive model;
• an act of reversing the transformations using the log to thereby change the second predictive model back to the first data compression predictive model;
• and
• an act of compressing a third segment of the data using a sequence of symbols that represents a third segment of the data, and also using the first data compression predictive model as a predictive starting point such that when compression of the third segment begins, the first data compression predictive model is based only on the first segment of data, such that when compression of the third segment of data begins, characters or symbols used for predicting the third segment of data are based entirely on the first predictive model of the first segment of data, and exclude consideration of the second segment of data and any intervening segments of data between the first segment of data and the third segment of data, and wherein the act of compressing the third segment of the data is performed after the act of compressing the first segment and after the act of compressing the second segment.

A computer program product for compressing data, the computer program product comprising one or more computer-readable physical storage media having stored thereon the following:

• computer-executable instructions for compressing a first segment of the data using a sequence of symbols that represents the first segment;
• computer-executable instructions for generating a first data compression predictive model which operates by predicting characters or symbols based on sequential processing of immediately preceding characters or symbols, and using the sequence of symbols that represents the first segment;
• computer-executable instructions for compressing a second segment of the data after compressing the first segment, using a sequence of symbols that represents a second segment of the data, and using the first data compression predictive model as a predictive starting point such that when compression of the second segment of data begins, such that when compression of the second segment of data begins, which is not sequential with the first segment of data, characters or symbols used for predicting the second segment of data are based entirely on the first predictive model of the first segment of data, and exclude consideration of all intervening segments of data between the first segment of data and the second segment of data, and wherein the first data compression predictive model is based only on the first segment of data;
• wherein the computer-executable instructions for compressing a second segment of the data using a sequence of symbols that represents a second segment of the data, and using the first data compression predictive model as a predictive starting point comprises the following:
• computer-executable instructions for compressing the first data compression predictive model while compressing the second segment to thereby generate a second predictive model after compressing the second segment;
• computer-executable instructions for generating a log that describes the transformations imposed on the first data compression predictive model as it changes to the second predictive model;
• computer-executable instructions for reversing the transformations using the log to thereby change the second predictive model back to the first data compression predictive model;
• and
• computer-executable instructions for compressing a third segment of the data after compressing the first segment, using a sequence of symbols that represents a third segment of the data, and also using the first data compression predictive model as a predictive starting point such that when compression of the third segment begins, the first data compression predictive model is based only on the first segment of data, such that when compression of the third segment of data begins, characters or symbols used for predicting the third segment of data are based entirely on the first predictive model of the first segment of data, and exclude consideration of the second segment of data and any intervening segments of data between the first segment of data and the third segment of data, and wherein the act of compressing the third segment of the data is performed after the act of compressing the first segment and after the act of compressing the second segment.

A method for decompressing segments of data as needed and without requiring sequential selection or decompression of the data even though the compression was performed using predictive models, the method comprising the following:

• an act of requesting at least a portion of an executable list;
• after requesting the at least a portion of the executable list, an act of receiving a compressed version of a first segment of the executable list over a network, the first segment being represented by a sequence of symbols and including one or more kernel functions that are the most common used when executing the executable list;
• an act of decompressing the first segment of the executable list, including the one or more kernel functions;
• an act of accessing a first data decompression predictive model which operates by predicting characters or symbols based on sequential processing of immediately preceding characters or symbols, wherein the first data decompression predictive model was generated using the sequence of symbols that represents the first segment;
• an act of determining that a second segment of the executable list is desirable, wherein the second segment does not appear sequentially with the first segment in the executable list and appears prior to the first segment in the sequential list, wherein the act of determining that a second segment of the data is desirable comprises the following:
• an act of executing the executable list within the first segment;
• while executing within the first segment, detecting that an instruction is needed from the second segment, wherein the act of detecting that an instruction is needed from the second segment comprises the following:
• an act of detecting an instruction in the first segment to go to an address in the second segment;
• an act of detecting the final instruction in the first segment;
• an act of accessing a compressed version of the second segment of the executable list;
• and
• an act of decompressing the second segment of the executable list after decompressing the first segment, and using the first data decompression predictive model to predict characters or symbols in the second segment of the executable list, wherein the first data decompression predictive model acts as a predictive starting point such that when decompression of the second segment of the executable list begins, the first data decompression predictive model is based only on the first segment of the executable list, such that when decompression of the second segment of the executable list begins, and which is not sequential with the first segment of the executable list, prediction of characters or symbols of the second segment of the executable list is based entirely on the first data decompression predictive model of the first segment of the executable list, and exclude consideration of all intervening segments of data between the first segment of the executable list and the second segment of the executable list, and wherein as the second segment of the executable list is decompressed, the first data decompression predictive model is progressed throughout sequential decompression of the second segment of the sequential list, such that all characters or symbols after the first character or symbol of the second segment are based only on the first segment of the executable list and preceding characters or symbols of the second segment of the executable list.