Background: Computer programming learning and education play a critical role in preparing a workforce equipped with the necessary skills for diverse fields. ChatGPT and YouTube are technologies that support self-directed programming learning. Objectives: This study aims to examine the sentiments and primary topics discussed in YouTube comments about ChatGPT's impact on learning and writing computer programming. Methods: The data were collected from 30 November 2022 to 11 January 2024, by extracting 30,773 comments from 57 YouTube videos. Sentiment analysis, topic modelling and thematic analysis were used for data analysis. Results and Conclusions: Through sentiment analysis and thematic analysis, a positive attitude among YouTube self-directed learners towards employing ChatGPT for learning and writing computer programming was identified. The results of topic modelling and thematic analysis revealed that these learners recognise both the perceived advantages and limitations of using ChatGPT for learning and writing computer programming. The advantages include creating learning plans, generating code, self-correction, explaining code and saving programming time, while the limitations are incorrect information, challenges in debugging programmes, perceived inefficiency and ineffectiveness and the absence of intelligence. Diverse perspectives regarding the impact of ChatGPT on programming professions and education were discussed. Some ethical concerns regarding data privacy, code copyright and equity issues were raised and needed further exploration. The findings imply the importance of computer programming education and integrating ChatGPT into programming education. Guidelines and instructions regarding using ChatGPT for programming learning are needed.